Re: [osol-discuss] Re: Sun to make Solaris more Linux like
a b wrote: Even if there is no VC money and some people will have to, hum, share an apartment? Why not? Remember also, VC capital is a U.S. specific thing. In Europe, if you want a startup, you fund that thing out of your own pocket. Off topic perhaps, but they have VC's in Britain as well now. Not as many as in the US, but some. Hugh. ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
This raises an interesting point. As a lot of comments regarding the lack of Linux documentation have been raised. The viewpoint of the Linux community is that OpenSolaris doesn't have enough documentation. Somehow we need to figure out what the disconnect is. The disconnect is that most newcomers have never heard of http://docs.sun.com/ Tomes and tomes of documentation have been written on just about every Solaris topic imaginable, and they are all on docs.sun.com. I'm under the impression, that when I redirect the a newb to docs.sun.com, they're just overwhelmed and don't want to read the docs, assuming beforehand that's it's going to be cut-and-dry materia, when in reality tons of documentation, especially Solaris documentation, are step-by-step guides with nice explanations of what one is doing with those steps. So next time somebody asks, redirect them to docs.sun.com. You'll be doing them a favor. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
I have started an ongoing personal project to engage my local Linux UG, (which I am a member of) in an OpenSolaris discussion. So far it's mostly been Where do I get and how do I install OpenSolaris?? Did you address the most common issues, such as: - software subsystem management commands (pkgadd, pkrgm, pkginfo) - Blastwave / pkg-get - bash versus tcsh, and why it's not necessarily a healthy idea to go around and change default settings (aka root's shell) - documentation, step-by-step HOWTOs on docs.sun.com - downloads on opensolaris.org - fastest changing codebase on the planet - OpenSolaris code base, Nevada distro, and Solaris 10 and 11? It would be interesting to read, what other questions pop up when you engage your local UG. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
IMHO, the Open Solaris community needs more than just programmers. Sure, but if someone that does documentation or marketing can code at least to the extent of the bite-size stuff, can in the former case read code without the need of constant consultation with the programmers, and in the latter case can comprehend documentation at least, then you don't have a bunch of disconnected functions all separately doing their things. They'd be capable of speaking about something to one another in common terms, even if when talking to consumers, they might use different language. Quite right. The traditional problem of marketing has been an apparent (and often frustrating) disconnect between the marketing people and technical understanding. I've myself sat on many a presentation of product that were potentially an excellent match for my company but the sale didn't go through because it became obvious very fast that the marketing guy (or girl) didn't really know the product beyond the layman abstraction and a few marketing terms. Marketing people have a tendency to want to abstract what they are presenting into layman's terms. This is fine and seems to work when one is selling a car, a toaster, or a washing machine. But for a highly technical, specialized market it is extremely frustrating! I want to know about the bucket you're trying to sell me, the makeup of the polymer chains and the strain factor of materials used. Meanwhile, the marketing guys keeps repeating that yes, it's a bucket, and it's a really great bucket that just has no equal in the market. But it wasn't a yes or no question! It was a concrete question on detail that will determine whether the bucket that guy is trying to sell to me will hold up when I fill it up with freshly mixed cement! Know your market. The managers might have the final say by signing the contract, but even the most non-technical manager has enough brains to at least probe the technical guys on their opinions before he signs that sales contract. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
RE: [osol-discuss] Re: Sun to make Solaris more Linux like
Even if there is no VC money and some people will have to, hum, share an apartment? Why not? Remember also, VC capital is a U.S. specific thing. In Europe, if you want a startup, you fund that thing out of your own pocket. Even if there is VC capital available, people here don't go for it, because if the firm here succeeds, the VCs will take it away from you, dress it up, then cash it in. So nobody here does it, because most people don't want their baby, that they've worked 16 hours a day for several years for, taken away by some bean counters. So yes, I still think the Solaris guy would be tempted, if he perceived that the idea was really solid. _ Invite your mail contacts to join your friends list with Windows Live Spaces. It's easy! http://spaces.live.com/spacesapi.aspx?wx_action=createwx_url=/friends.aspxmkt=en-us___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
There's lots of that, but the opposite is also true: If you wait for the VCs to plunk down $3,000,000 before you get your app up, and if you wait until the Solaris Guy believes in your little group to join (he won't join until the VC money is in, the VC won't give you the money until he's on board, king of thing), then you end up with a great dream, and not much else. There's a lot of that going on. Let me tell you a true story: A Solaris guy will never join a rogue Linux startup. First of all, the developers will consider him to be obsolete - simply WORTHLESS. No, lower than worthless. If a Solaris guy does join this startup, it will not be because the developers there have enough brains to perceive that he can do a whole lot more than a Linux guy could, but because that Solaris guy has a whole bunch of unseen-before Linux experience, like running anything from TurboLinux on Itanium to RedHat 4 AS, which will make these geeks drool. Sad thing is, they will be drooling for a completely wrong reason. And if the Solaris guy does accept the job, he will be totally and completely miserable in such a shop. First, he will inherit a total and complete MESS, composed of a mashup of systems that can never be made to work right. He'll have to fight, day in and day out, babysitting these Linux systems, a nightmare for every self-respecting sysadmin, and nothing short of catastrophe for a good system engineer. Now, when this poor guy tries to actually do something, like put Solaris in to put an end to this insanity, he'll be told that the firm is really big on Linux and that that's what the future is for this company. So right down the toilet goes any chance of designing and implementing a rock solid infrastructure. He'll be forbidden to put any Solaris in because the developers will be freaking out in fear from the big bad unknown Solaris. Not even offering to teach them UNIX on a regular basis will help qualm their fear of the unknown. Eventually, somehow, the business will take off. Except for a small problem, as soon as the business gets critical mass, the firm will start getting big customers. And all of a sudden, customers start demanding solutions on Solaris, and don't want to hear anything about Linux. Even though a Solaris infrastructure is now absolutely required, the management will deny all and any requests to put in any Solaris systems. So much so, that the firm will be reduced to begging their own customers to LEND THEM some Solaris systems, so that the firm could develop the software for the customer. Oh, and let's not forget, even though the Solaris guy will be expected to keep everything running smoothly, no way in a thousand hells will he ever see, let alone get one penny from those $3,000,000 venture capital. He'll just be expected to magically pull solutions and fix broken, hacked-together hardware with funds he pulls out of his behind. Like I wrote at the beginning, this is a true story. So no worries about Solaris guys joining Linux startups - it's *extremely unlikely*. In corporate speak, we don't really have a match in profiles. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Let me tell you a true story: Replace Solaris guy with debian guy in a Redhat shop. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
--- UNIX admin [EMAIL PROTECTED] wrote: Let me tell you a true story: A Solaris guy will never join a rogue Linux startup. First of all, the developers will consider him to be obsolete - simply WORTHLESS. No, lower than worthless. Sad story. But let's say the developers of the idea are python+framework+ui+graphics people, and they are os agnostic, and one of them says: we should get solaris because it'll scale, and if we can bring in Joe the Solaris Guru at founding and give him shares and make him Director of OS infrastructure... Sounds like the founders have their act together, no? But can they get the Solaris guy to leave his $120,000 corporate data-center multi-thousand server job? No. So they flounder around a bit and eventually get a University student who uses Gentoo who's all fired up. They get Linux in, and then they build their web 2.0 application, get it on Digg and have 50,000 teenagers go nuts over it. Then they suffer, and Mr gentoo returns to school full time. Then, they look at Solaris. But it's too late to learn: they're dropping connections left and right and the fickle public has moved on to the next act. Chris Mahan 818.943.1850 cell [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.christophermahan.com/ Take the Internet to Go: Yahoo!Go puts the Internet in your pocket: mail, news, photos more. http://mobile.yahoo.com/go?refer=1GNXIC ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
RE: [osol-discuss] Re: Sun to make Solaris more Linux like
--- a b [EMAIL PROTECTED] wrote: Sad story. But let's say the developers of the idea are python+framework+ui+graphics people, and they are os agnostic, and one of them says: we should get solaris because it'll scale, and if we can bring in Joe the Solaris Guru at founding and give him shares and make him Director of OS infrastructure... Sounds like the founders have their act together, no? But can they get the Solaris guy to leave his $120,000 corporate data-center multi-thousand server job? No. So they flounder around a bit and eventually get a University student who uses Gentoo who's all fired up. What you describe is almost too good to be true. If that were indeed the case, I think that the Solaris guy might seriously be tempted. Data centers run themselves. Sure there is always work to do, but it's all incremental. Who could say no to such an opportunity, to build stuff from the ground up once again, with all the errors from previous experience, corrected? It would be a tough thing to say no to. Remember that every good system engineer is deeply passionate about technology. Even if there is no VC money and some people will have to, hum, share an apartment? Chris Mahan 818.943.1850 cell [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.christophermahan.com/ Moody friends. Drama queens. Your life? Nope! - their life, your story. Play Sims Stories at Yahoo! Games. http://sims.yahoo.com/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
RE: [osol-discuss] Re: Sun to make Solaris more Linux like
Sad story. But let's say the developers of the idea are python+framework+ui+graphics people, and they are os agnostic, and one of them says: we should get solaris because it'll scale, and if we can bring in Joe the Solaris Guru at founding and give him shares and make him Director of OS infrastructure... Sounds like the founders have their act together, no? But can they get the Solaris guy to leave his $120,000 corporate data-center multi-thousand server job? No. So they flounder around a bit and eventually get a University student who uses Gentoo who's all fired up. What you describe is almost too good to be true. If that were indeed the case, I think that the Solaris guy might seriously be tempted. Data centers run themselves. Sure there is always work to do, but it's all incremental. Who could say no to such an opportunity, to build stuff from the ground up once again, with all the errors from previous experience, corrected? It would be a tough thing to say no to. Remember that every good system engineer is deeply passionate about technology. _ News, entertainment and everything you care about at Live.com. Get it now! http://www.live.com/getstarted.aspx___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
On 5/22/07, Eric Boutilier [EMAIL PROTECTED] wrote: The design and planning of this (an OpenSolaris reference distro) _is_ going to be an OpenSolaris Community-governed thing, not a Sun-governed thing, right? Yes. -ian -- Ian Murdock 650-331-9324 http://ianmurdock.com/ Don't look back--something might be gaining on you. --Satchel Paige ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
--- Chung Hang Christopher Chan [EMAIL PROTECTED] wrote: Let me tell you a true story: Replace Solaris guy with debian guy in a Redhat shop. I meant there is a another true story like this only with a debian guy in a redhat shop. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
On Mon, 21 May 2007, Ian Murdock wrote: On 5/21/07, Brian Gupta [EMAIL PROTECTED] wrote: I don't think one distro can be all things to all people. Agreed. That's why there's still plenty of room for other distros, regardless of what Sun does. Ian, you aren't gonna win this one by trying to convince everyone here that the Linux way is the way. Don't talk about it till you are blue in the face, as there are certain people you will never convince.. I suggest doing an end around, and just make it happen. Gather together a few like minded devs, and just do it. Don't worry, I know how this works. I suspect I've convinced everyone I'm going to convince at this point, which is enough to start moving the ball forward, and I also understand there's a certain contingent I'm never going to convince no matter what I say or do. We'll be setting the formal machinery in motion later in the week to make this actually happen... Who's We? Sorry, but that comes off as if you just made a nasty Freudian slip. Which then begs the question again: The design and planning of this (an OpenSolaris reference distro) _is_ going to be an OpenSolaris Community-governed thing, not a Sun-governed thing, right? Eric -ian -- Ian Murdock 650-331-9324 http://ianmurdock.com/ Don't look back--something might be gaining on you. --Satchel Paige ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
On 5/22/07, Eric Boutilier [EMAIL PROTECTED] wrote: On Mon, 21 May 2007, Ian Murdock wrote: On 5/21/07, Brian Gupta [EMAIL PROTECTED] wrote: I don't think one distro can be all things to all people. Agreed. That's why there's still plenty of room for other distros, regardless of what Sun does. Ian, you aren't gonna win this one by trying to convince everyone here that the Linux way is the way. Don't talk about it till you are blue in the face, as there are certain people you will never convince.. I suggest doing an end around, and just make it happen. Gather together a few like minded devs, and just do it. Don't worry, I know how this works. I suspect I've convinced everyone I'm going to convince at this point, which is enough to start moving the ball forward, and I also understand there's a certain contingent I'm never going to convince no matter what I say or do. We'll be setting the formal machinery in motion later in the week to make this actually happen... Who's We? Sorry, but that comes off as if you just made a nasty Freudian slip. Which then begs the question again: The design and planning of this (an OpenSolaris reference distro) _is_ going to be an OpenSolaris Community-governed thing, not a Sun-governed thing, right? We = me + the people at Sun who are working with me--in this case, the people figuring out the necessary buttons to push and necessary levers to pull to spin up the appropriate body in the OpenSolaris community to do the design and planning (and implementation). There is no conspiracy.. -ian -- Ian Murdock 650-331-9324 http://ianmurdock.com/ Don't look back--something might be gaining on you. --Satchel Paige ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
On Tue, 22 May 2007, Ian Murdock wrote: On 5/22/07, Eric Boutilier [EMAIL PROTECTED] wrote: On Mon, 21 May 2007, Ian Murdock wrote: On 5/21/07, Brian Gupta [EMAIL PROTECTED] wrote: I don't think one distro can be all things to all people. Agreed. That's why there's still plenty of room for other distros, regardless of what Sun does. Ian, you aren't gonna win this one by trying to convince everyone here that the Linux way is the way. Don't talk about it till you are blue in the face, as there are certain people you will never convince.. I suggest doing an end around, and just make it happen. Gather together a few like minded devs, and just do it. Don't worry, I know how this works. I suspect I've convinced everyone I'm going to convince at this point, which is enough to start moving the ball forward, and I also understand there's a certain contingent I'm never going to convince no matter what I say or do. We'll be setting the formal machinery in motion later in the week to make this actually happen... Who's We? Sorry, but that comes off as if you just made a nasty Freudian slip. Which then begs the question again: The design and planning of this (an OpenSolaris reference distro) _is_ going to be an OpenSolaris Community-governed thing, not a Sun-governed thing, right? We = me + the people at Sun who are working with me--in this case, the people figuring out the necessary buttons to push and necessary levers to pull to spin up the appropriate body in the OpenSolaris community to do the design and planning (and implementation). There is no conspiracy.. You answered as if I asked Is this a conspiracy? (Which, by the way, is an untactful inference which I take exception to. But maybe, in similar fashion, you took exception to Freudian slip. So, touche, I guess). What I actually did ask was this: The design and planning of this (an OpenSolaris reference distro) _is_ going to be an OpenSolaris Community-governed thing, not a Sun-governed thing, right? Eric ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
--- Eric Boutilier [EMAIL PROTECTED] wrote: The design and planning of this (an OpenSolaris reference distro) _is_ going to be an OpenSolaris Community-governed thing, not a Sun-governed thing, right? Maybe Sun got tired of waiting for the Community? Chris Mahan 818.943.1850 cell [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.christophermahan.com/ Fussy? Opinionated? Impossible to please? Perfect. Join Yahoo!'s user panel and lay it on us. http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7 ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
I read the we as everyone [I've convinced] at this point. Also, it's been said over and over that it will be a community project. To me that implies that it will be mostly Sun people, because Sun is most of the active community :) This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
. . . snip . . . There's lots of that, but the opposite is also true: If you wait for the VCs to plunk down $3,000,000 before you get your app up, and if you wait until the Solaris Guy believes in your little group to join (he won't join until the VC money is in, the VC won't give you the money until he's on board, king of thing), then you end up with a great dream, and not much else. There's a lot of that going on. The alternative is to use whatever you know, be it Debian in my case, or ubuntu, Suse, Fedora, whatever, and getting it done. It's better to try and fail than not try at all. At least, if you fail spectacularly, and not because of a bad idea but because of poor hardware/software choices, you have a better shot with the VCs next time around. Agreed. One of most effective approaches to promote OpenSolaris is to think in terms of VCs, i.e., how a VC (especially those preemptive microVCs) may be interested in a Solaris-related project that is not connected to any establishment including Sun. . . . snip . . . A coworker of mine once argued that bittorent would have been much more better/robust/whatever written in java rather than in python. I replied: That may be true, but it wasn't written in java, was it? Azureus is arguably the best and also the most robust BitTorrent client, which, incidentally, happens to be written in java. However, it is not available in many Linux distros including Debian, Fedora/RedHat, etc., unless you specifically install it. Things are rapidly changing now that java is being GPL'ed. . . . snip . . . Remember: if it can't run on $400 hardware (typical repurposed corporate pc), they won't look at it. Solaris runs quite comfortably on an Athlon64/mobo combo that costs $69.99 USD: http://promotions.newegg.com/AMD/939pin/index.html?CMP=EMC-IGNEFL052207ATT=AMDFox-Combo There is little excuse not to seriously consider Solaris as the primary OS for enterprises. . . . snip . . . Chris Mahan 818.943.1850 cell [EMAIL PROTECTED] [EMAIL PROTECTED] http://www.christophermahan.com/ This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
No CIO in the world said, I've gotta get me some Linux!, or at least in the late 1990s when Linux was taking off. He woke up one day and realized Linux was already everywhere. Actually the reality is that he was told some day that Linux has already been put in place by some bloke on cheapo DELL hardware, after the fact. Number two thing that happened, InfoWeek and all those management garbage magazines got on the bandwagon - Linux was just the sensationalist thing to write about and keep the subscriptions going. The fact that most managers are economists and not technical people didn't help either. Chanting something long enough to onself or reading the same thing over and over - and perception becomes reality. Meanwhile neither DELL nor Linux are cheap any more - and Fedora, even though it has an initial price tag of gratis - is extremely expensive in the long run. Does the economist know that? No. Does the hacker who put it in know? No. Do the hacker or the economist know any better? Again, the answer is a resounding no, or they wouldn't have done it / approved it. Take a wild guess what will happen when the hacker finally gets bored or when the economist pisses him off enough so that he leaves? BTW, ever heard of CMM, the Capability Maturity Model? Do you fight the trend or figure out how to take advantage of it? The answer seems perfectly logical to me. Not at the expense of quality! You want to take advantage of the trend? Then use the perception is reality method. Fight fire with fire. Educate the user base. Reality is, Solaris is easy. Much easier than Linux. Perception tells us the opposite. We need to change the perception, then the reality will change also. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 5/17/07, James Carlson [EMAIL PROTECTED] wrote: Ian Murdock writes: On 5/17/07, Richard L. Hamilton [EMAIL PROTECTED] wrote: Until an open collaboration of developers achieves the same documented process as industry or formal membership based bodies (ECMA, IEEE) or national (ANSI) or international standards bodies (ISO), _and_ there is a commitment to developing their standards in such a way that they are more or less platform neutral (which ultimately requires multi-platform development, i.e. participation of competing interests), I don't see how a group of developers can constitute any sort of a standard; at most, the product of their work will become the winner of an informal popularity contest. IETF? There are some notable outlier cases, but the IETF primarily sets standards for the bits on the wire. In general, it's not at all concerned about user interface, operating system, or application design -- or even with coding -- which are exactly the issues we're discussing here. I don't see how that's relevant for the question at hand, which I believe was the importance of POSIX, SUS, and similar OS standards for modern systems (or the lack thereof). It's all interfaces at the end of the day, isn't it? I.e., making sure the receiver understands what the sender is trying to tell it. That may take the form of bits on the wire, or library functions, or something else, but it's all shades of the same thing. In that context IETF shows that standards, effective standards, can be built bottom-up--it doesn't have to be a top-down world, which was the original assertion. -ian -- Ian Murdock 650-331-9324 http://ianmurdock.com/ Don't look back--something might be gaining on you. --Satchel Paige ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
RE: [osol-discuss] Re: Sun to make Solaris more Linux like
Sure, IBM may be not be innovating with AIX, as we speak, they've certainly done so in the past, and it would be a shame to ignore that. Solaris has gone through its dark times, as well, when AIX was considered innovative (consider, Solaris 8 v. AIX 4.3.3 or Solaris 9 v. AIX 5.2, which is even more of a gap). They all have kept up with each other over the years; it was almost like a social club. One of the examples I like to take as a typical is installation systems: Solaris comes out with JumpStart(TM) HP-UX comes out with Ignite-UX(TM) IRIX suddenly gets roboinst They are all implemented and configured differently, but they essentially do the same thing. It was always like this. What one had, the other one got also, in one form or another, sooner or later. If one just stopped and listened and looked, one could see them keeping pace with one another. But IRIX is now dead HP-UX is just in minimum keep up mode what do we hear about innovations in AIX? Can't say I've seen IBM divluge anything of any significance. Solaris is the only one slamming the pedal to the metal, unlike anything I've ever seen before. And I sure hope that that V10 engine holds up at such high RPMs. _ Explore the seven wonders of the world http://search.msn.com/results.aspx?q=7+wonders+worldmkt=en-USform=QBRE___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Ian Murdock writes: On 5/17/07, James Carlson [EMAIL PROTECTED] wrote: I don't see how that's relevant for the question at hand, which I believe was the importance of POSIX, SUS, and similar OS standards for modern systems (or the lack thereof). It's all interfaces at the end of the day, isn't it? More or less, yes. I.e., making sure the receiver understands what the sender is trying to tell it. That may take the form of bits on the wire, or library functions, or something else, but it's all shades of the same thing. For architecture on a given system, yes, there's commonality here. For standards setting purposes, there isn't. The key issue with respect to the IETF is that the focus is on internetworking protocols. The operating system and its constraints are largely irrelevant. This is a very important point, because the horizons involved are very different. From the IETF point of view, operating systems come and go frequently. Protocols are forever. The people who gather to work on IETF proposals are generally the ones implementing wire protocols. Thus, if you take an API proposal to them, they won't necessarily be the right people to evaluate it. Not only will you be missing domain experts from applications areas and particular operating systems, but the people looking at it are often working in a different area of their respective systems. In that context IETF shows that standards, effective standards, can be built bottom-up--it doesn't have to be a top-down world, which was the original assertion. I don't think I see it that way at all. Neither one is any more top down (would you call POSIX top down?) or bottom up than the other. The two worlds (de jure standards groups, such as ISO, and IETF) are quite different. I think I know what you're saying -- noting the fact that the IETF isn't a membership-based organization and that it doesn't have representatives like those other groups -- but I disagree that these characteristics necessarily make it a good place to set programming interface standards. And I would very much disagree with an implication that we should start ignoring one class of standards because we think another class is coming into favor. -- James Carlson, Solaris Networking [EMAIL PROTECTED] Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Chanting something long enough to onself or reading the same thing over and over - and perception becomes reality. Interesting that you should say that... :) This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 5/21/07, UNIX admin [EMAIL PROTECTED] wrote: No CIO in the world said, I've gotta get me some Linux!, or at least in the late 1990s when Linux was taking off. He woke up one day and realized Linux was already everywhere. Actually the reality is that he was told some day that Linux has already been put in place by some bloke on cheapo DELL hardware, after the fact. Isn't that what I said? Chanting something long enough to onself or reading the same thing over and over - and perception becomes reality. Heheh. Meanwhile neither DELL nor Linux are cheap any more - and Fedora, even though it has an initial price tag of gratis - is extremely expensive in the long run. Agreed. That's what I'm calling this an OPPORTUNITY for OpenSolaris. -ian -- Ian Murdock 650-331-9324 http://ianmurdock.com/ Don't look back--something might be gaining on you. --Satchel Paige ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Chanting something long enough to onself or reading the same thing over and over - and perception becomes reality. Interesting that you should say that... :) I've thought and considered these things for a very long time. I still think about them every day. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
RE: [osol-discuss] Re: Sun to make Solaris more Linux like
Do you have a HP-UX-11.x system? Yes, I do. Well, there is AIX but I am not sure whether IBM takes it for real and I know of no hacker who is using AIX as development platform. I wanted to get AIX, but when I looked even at an outdated 32-bit rack-mountable AIX PPC system, the prices of the hardware were so high I just said - forget it! And the fact that one can't get AIX readily makes things even worse. Perhaps it is possible to obtain AIX gratis, but it certainly isn't obvious how and requires further digging, however, without IBM's PPC hardware to run it on, it's pointless. I know I could get an old slow RS/6000 system, but I only buy 19 rack-mountable stuff. It's really not too bright to buy hopelessly obsolete HW (old but useful is fine, as long as it's rack-mountable). Plus, I've tracking AIX on and off, and I really don't see any significant innovation being done on it. IBM does the bare minimum to keep the platform so-so with the times... but IBM sure seems to do it with grinding their teeth, only when they must. That to me is no innovation, but idiocy. They can keep cowering before the sitting penguin for all I care, they're useless weaklings. Sun is very happy with the easy success in east Europe but conentrating to east Europe only is the wrong way. To be fair, parts of eastern (actually central, but *falsely* labeled eastern) Europe, like Croatia, have always been Solaris strongholds, and I personally know a few top-notch Solaris experts coming from there, that are now all over the world. They have CARNet (Croatian Academic and Research Network) to thank for for that. _ Discover the new Windows Vista http://search.msn.com/results.aspx?q=windows+vistamkt=en-USform=QBRE___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
a b wrote: I wanted to get AIX, but when I looked even at an outdated 32-bit rack-mountable AIX PPC system, the prices of the hardware were so high I just said - forget it! And the fact that one can't get AIX readily makes things even worse. Perhaps it is possible to obtain AIX gratis, but it certainly isn't obvious how and requires further digging, however, without IBM's PPC hardware to run it on, it's pointless. One of the features of AIX and the pSeries hardware is the strength of the relationship between the two. AIX on commodity hardware would be quite pointless, and I always tell those this that want to run AIX on commodity hardware. The amount of awareness AIX has with regard to the underlying hardware is quite amazing, and is certainly comparable to Solaris on SPARC if not more. I purchased an IBM pSeries p640 (also known as an RS/6000 7026-B80) off of eBay a little over a year ago for $1,200. RS/6000 hardware can be had for cheaper, but this system has 2x375MHz POWER3-IIs and 2GB of memory. Performance is roughly comparable to a 750MHz UltraSPARC-III, as such systems were sold about the same time as the p640 (~2000). You can purchase its single-proc variant (the p640 is quad-capable) for $300-$400, nowadays. Plus, I've tracking AIX on and off, and I really don't see any significant innovation being done on it. IBM does the bare minimum to keep the platform so-so with the times... but IBM sure seems to do it with grinding their teeth, only when they must. I wouldn't quite say AIX isn't innovating and its definitely not the case it hasn't ever innovated. AIX and Solaris have often benefited from each other. Examples of this would be: (1) AIX moved from the M:N threading model before Solaris did (2) Solaris had Live Upgrade before AIX did, but AIX now has it, as well (implemented in 5.2, IIRC) (3) AIX and Solaris both have pageable kernel spaces to some extent (AIX seems to be able to page more of the the kernel space than Solaris from what I understand (4) for years AIX had a better volume manager that was included with the operating system. Even with the amount of suckage SVM has, it was an add-on up until not too long ago, while the LVM in AIX has always been coupled with the OS base. (5) AIX had tracing capabilities, though primitively, years before Solaris did (6) some would say pSeries/AIX has far better hardware virtualization than domains on Sun hardware Sure, IBM may be not be innovating with AIX, as we speak, they've certainly done so in the past, and it would be a shame to ignore that. Solaris has gone through its dark times, as well, when AIX was considered innovative (consider, Solaris 8 v. AIX 4.3.3 or Solaris 9 v. AIX 5.2, which is even more of a gap). I administer and work with Solaris and Sun hardware far more than pSeries/AIX, and would certainly recommend going with Solaris to a client or customer over AIX, but ignoring how AIX and Solaris have followed each other's developments is silly, IMO. Derek E. Lewis [EMAIL PROTECTED] http://delewis.blogspot.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Alan DuBoff wrote: On Fri, 18 May 2007, Jim Grisanzio wrote: Sun is actually a massive a minority in the community at this point. There are 51,179 people registered on the site right now, and only about 2,000 of those have Sun badges. So, where are all those other people? I don't see them participating on the mailing lists. Some have been signing up all along, of course, but some have registered due to the Starter Kit program (a lot, actually) recently, and many are now coming to OpenSolaris via conferences. Just in the last year or so OpenSolaris community members have been to Tech Days (about 12 cities globally), ApacheCon, OSCON, EuroOSCON, Japan Open Source Conf, Germany OpenSolaris Conf, FOSDEM, Colorado Tech/Ed Conf, China Software Summit, LinuxWorld, JavaOne ... and probably a few others I can't think of. Not to mention the UG out there, too. Now that we are going out and talking, people are starting to come to us. This is hugely positive. The next step is to help them get more directly involved in the community. And they are, actually. Mail list activity and web forum hits have grown consistently since we launched, and the rate of grow is increasing. We are starting to put some pretty big numbers on the board, actually. There are about 200 lists now, so you may not see all the increase in activity on this list since it's widely distributed now. Also keep in mind that there are language and culture barriers we need to overcome as we grow outside the west. That remains an untapped opportunity, and one that is very exciting. It's been great to see the UG lists diversify into various languages as communities develop in far away places -- but they /are/ connected to opensolaris.org and they /want to be/ connected to opensolaris.org. We have to accept the fact that all the growth in the community will not take place on opensolaris-discuss, or in California, or even in the U.S. for that matter. The community will look different in different regions. I actually think the real challenge we have is not in building one monolithic OpenSolaris community but in connecting many (sometimes disparate) OpenSolaris communities. Those are marketing numbers, they mean nothing to what is done. They are opensolaris.org registration numbers and they supported by OpenSolaris engineering, OpenSolaris marketing, and other groups that are participating in promoting OpenSolaris around the world. They represent the good work of many people who are part of this community. This is no different than how Sun counts the downloads. Sure, the OGB is primarily Sun people, but that's who we elected. I doubt that situation will be the same in the future, but ultimately, it's up to community members to get involved and participate. Absolutely, and I agree it will change in the future. But until then the communities are dominated with Sun people and it's hard for anyone to have a pulse on each community without being subscribed. But you can't get to the until then without taking some steps first. After all, you construct a building by digging a hole first. We are building all the necessary things a community would need to function -- code, tools, infrastructure, governance, process, people, promotion, whatever. Buildings don't go up over night, and communities are not built over night, either. We need to understand that it exists how it is today. It is a fact that Sun has the largest influence with OpenSolaris as it is today. They will also continue to have a large influence for the future as well. When you say Sun what do you mean? Just because there are many Sun employees involved in certain aspects of the project is not a problem. Actually, we need /more/ Sun people involved as the core SCM infrastructure migrates external. From an engineering standpoint, this is good. From other areas of Sun, I don't know. I'm confused. Sun should control some parts but not others? OpenSolaris should be about open development (engineering) and community building (like what you are doing with the SVOSUG), not about products and services and competitive issues with other companies and communities. I agree with this statement, but let me explain something to you about SVOSUG and how I run it. There is very little input from Sun on what I should present, or how it should be presented. To date, I have looked at technologies that are happening in Solaris and lined up speakers to talk about them. I have not asked Sun for much in the way of funding, other than meeting space. Flip Russell provides drinks which are funded by Sun, but that is minimal. And I think that's great. I also think your group is one of the most successful. I have run many user groups, and been involved in many more. What I have learned is that these user groups are best grown out of grass roots, because it's the people who show up and are there to meet and talk with others
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Doug Scott schrieb: The patch for df.c below checks to see if stdout is a tty (i.e. interactive), and will add the -h option if no other options conflict. If stdout is not a tty (i.e. a script), the df command is unchanged. Is there anything wrong with this idea?? Why does everyone like the -h output? I personally hate it - and would hate it even more if it would be the default. It probably shouldn't be the default, but I can't really tell what this means: 131996712968156 9652530272 1% quick, how much space is that? With the -h option the width of the filesystem size is roughly the same - regardless if the filesystem is 1kB or 1TB. So I have to actually *read* the output instead of a quick look how wide the size column is. That only helps upto about 6-8 digits, if that. Casper ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
a b [EMAIL PROTECTED] wrote: Do you have a HP-UX-11.x system? Yes, I do. Well, there is AIX but I am not sure whether IBM takes it for real and I know of no hacker who is using AIX as development platform. I wanted to get AIX, but when I looked even at an outdated 32-bit rack-mountable AIX PPC system, the prices of the hardware were so high I just said - forget it! And the fact that one can't get AIX readily makes things even worse. Perhaps it is possible to obtain AIX gratis, but it certainly isn't obvious how and requires further digging, however, without IBM's PPC hardware to run it on, it's pointless. I know I could get an old slow RS/6000 system, but I only buy 19 rack-mountable stuff. It's really not too bright to buy hopelessly obsolete HW (old but useful is fine, as long as it's rack-mountable). It seems that all refurbished AIX systems that include a newer version of the OS are illegal. I tried to get a machine for a porting project together with a license document to no avail. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
[EMAIL PROTECTED] schrieb: It probably shouldn't be the default, but I can't really tell what this means: 131996712968156 9652530272 1% quick, how much space is that? Enough free. With the -h option the width of the filesystem size is roughly the same - regardless if the filesystem is 1kB or 1TB. So I have to actually *read* the output instead of a quick look how wide the size column is. That only helps upto about 6-8 digits, if that. The problem is the formatting. In Solaris df/ls the size column has a fixed width. For larger sizes the output gets distorted: -rwsr-s--x 1 oracle dba 133894000 Mai 29 2006 oracle* -rwxr-xr-x 1 oracle dba 1339524 Mai 19 2006 oratclsh* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxml* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxsl* -rwxr-xr-x 1 oracle dba 10636376 Mai 29 2006 proc* FreeBSD's utilities in contrast check for the largest required width and adjusts output accordingly: -rwsr-s--x 1 oracle dba 133894000 Mai 29 2006 oracle* -rwxr-xr-x 1 oracle dba1339524 Mai 19 2006 oratclsh* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxml* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxsl* -rwxr-xr-x 1 oracle dba 10636376 Mai 29 2006 proc* So even If you don't see the absolute numbers you quickly realize that oracle is two magnitudes larger than oratclsh. If df had a similar output, I'd prefer this: 83886083041200 534740837% 335544321605719 31794110 5% 62914560 37766679 1412994873% 62914560 11017932 1412994844% 1535901696 1187247048 26307973582% 134217728 32200480 7472805031% 41943041392793 280151134% over this any time: 8.0G 2.9G 5.1G37% 32G 1.5G30G 5% 60G36G13G73% 60G11G13G44% 1.4T 1.1T 251G82% 128G31G71G31% 4.0G 1.3G 2.7G34% I don't count digits. I just memorize patterns. More digits = larger size. Daniel ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
FreeBSD's utilities in contrast check for the largest required width and adjusts output accordingly: -rwsr-s--x 1 oracle dba 133894000 Mai 29 2006 oracle* -rwxr-xr-x 1 oracle dba1339524 Mai 19 2006 oratclsh* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxml* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxsl* -rwxr-xr-x 1 oracle dba 10636376 Mai 29 2006 proc* There's this thing with cut which makes such output problematic. So even If you don't see the absolute numbers you quickly realize that oracle is two magnitudes larger than oratclsh. If df had a similar output, I'd prefer this: 83886083041200 534740837% 335544321605719 31794110 5% 62914560 37766679 1412994873% 62914560 11017932 1412994844% 1535901696 1187247048 26307973582% 134217728 32200480 7472805031% 41943041392793 280151134% over this any time: 8.0G 2.9G 5.1G37% 32G 1.5G30G 5% 60G36G13G73% 60G11G13G44% 1.4T 1.1T 251G82% 128G31G71G31% 4.0G 1.3G 2.7G34% I don't count digits. I just memorize patterns. More digits = larger size. Sometimes it's nice to know what the size is when you want to do something large. I must admit, though, that I do not like the way we currently split these values; e.g., I prefer 1438G over 1.4T (loss of precision, one space sacrificed to a useless .. Similarly for 2713M over 2.7G Casper ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
[EMAIL PROTECTED] schrieb: FreeBSD's utilities in contrast check for the largest required width and adjusts output accordingly: -rwsr-s--x 1 oracle dba 133894000 Mai 29 2006 oracle* -rwxr-xr-x 1 oracle dba1339524 Mai 19 2006 oratclsh* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxml* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxsl* -rwxr-xr-x 1 oracle dba 10636376 Mai 29 2006 proc* There's this thing with cut which makes such output problematic. Why? In the example above with cut (and absolute columns) you also end up in the wrong column if the file size is large enough (10MB and larger). Daniel ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Daniel Rock [EMAIL PROTECTED] wrote: If df had a similar output, I'd prefer this: 83886083041200 534740837% 335544321605719 31794110 5% 62914560 37766679 1412994873% 62914560 11017932 1412994844% 1535901696 1187247048 26307973582% 134217728 32200480 7472805031% 41943041392793 280151134% over this any time: 8.0G 2.9G 5.1G37% 32G 1.5G30G 5% 60G36G13G73% 60G11G13G44% 1.4T 1.1T 251G82% 128G31G71G31% 4.0G 1.3G 2.7G34% I don't count digits. I just memorize patterns. More digits = larger size. I only use the ls -h output in case I need to know the amount. In most cased, I prefer the normal ls output. Note that I frequently use: ls -l | sort -n +4 to find the largest file and this will not work with ls -lh. For df, I use df -h more frequently but I still don'e like df -h to be the default. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Compared to other marketing activities from Sun, this would be cheap and the current idea of project Indiana looks to me like a Sun OpenSolaris distribution that (if done the way it currently seems) will most likely embrace and crush the sensitive plants that are the real free grown distributions. I do not see why a Sun OpenSolaris distribution will kill off your 'real free' distribution or any of the other 'real free' distributions. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Chung Hang Christopher Chan [EMAIL PROTECTED] wrote: Compared to other marketing activities from Sun, this would be cheap and the current idea of project Indiana looks to me like a Sun OpenSolaris distribution that (if done the way it currently seems) will most likely embrace and crush the sensitive plants that are the real free grown distributions. I do not see why a Sun OpenSolaris distribution will kill off your 'real free' distribution or any of the other 'real free' distributions. I don't see that it would help to have a second Sun Solaris distribution. If Sun likes to put money into OpenSolaris, this should be done in a way that enables collaboration and in a way that allows to contribute code by non-Sun people. Sun has the best OSS concept compared to other companies (e.g. Apple) but a concept is not sufficient, it needs tp be turned into reality. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Chung Hang Christopher Chan [EMAIL PROTECTED] wrote: Compared to other marketing activities from Sun, this would be cheap and the current idea of project Indiana looks to me like a Sun OpenSolaris distribution that (if done the way it currently seems) will most likely embrace and crush the sensitive plants that are the real free grown distributions. I do not see why a Sun OpenSolaris distribution will kill off your 'real free' distribution or any of the other 'real free' distributions. I don't see that it would help to have a second Sun Solaris distribution. If Sun likes to put money into OpenSolaris, this should be done in a way that enables collaboration and in a way that allows to contribute code by non-Sun people. It doesn't matter if it helps or not, or does it? If Sun feels justified in spending its money that way, then why can it not spend it that way. Casper ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
--- Joerg Schilling [EMAIL PROTECTED] wrote: Chung Hang Christopher Chan [EMAIL PROTECTED] wrote: Compared to other marketing activities from Sun, this would be cheap and the current idea of project Indiana looks to me like a Sun OpenSolaris distribution that (if done the way it currently seems) will most likely embrace and crush the sensitive plants that are the real free grown distributions. I do not see why a Sun OpenSolaris distribution will kill off your 'real free' distribution or any of the other 'real free' distributions. I don't see that it would help to have a second Sun Solaris distribution. If Sun likes to put money into OpenSolaris, this should be done in a way that enables collaboration and in a way that allows to contribute code by non-Sun people. What makes you think that will not happen? I see a Sun distribution that is different from Solaris 10 as useful in getting current Linux users to switch as opposed to waiting for a new generation of Solaris users from universities. Especially since they will first need to get a job that involves using Solaris whereas current Linux users can switch their systems over if the OpenSolaris distribution does not present too much of a fear of the unknown. Sun has the best OSS concept compared to other companies (e.g. Apple) but a concept is not sufficient, it needs tp be turned into reality. Agreed. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
[EMAIL PROTECTED] wrote: FreeBSD's utilities in contrast check for the largest required width and adjusts output accordingly: -rwsr-s--x 1 oracle dba 133894000 Mai 29 2006 oracle* -rwxr-xr-x 1 oracle dba1339524 Mai 19 2006 oratclsh* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxml* -rwxr-xr-x 1 oracle dba 48 Sep 25 2000 oraxsl* -rwxr-xr-x 1 oracle dba 10636376 Mai 29 2006 proc* There's this thing with cut which makes such output problematic. You can use (n)awk in the case above. For example, what happens with cut in the case above if a future project allows usernames 8 characters? But granted that some people will have scripts that use cut, so it's hard to change. Hugh. ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Hugh McIntyre schrieb: You can use (n)awk in the case above. For example, what happens with cut in the case above if a future project allows usernames 8 characters? But granted that some people will have scripts that use cut, so it's hard to change. Compatible formatting if stdout is not a terminal. Optimized formatting if stdout is a terminal. Daniel ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 19/05/07, Daniel Rock [EMAIL PROTECTED] wrote: Hugh McIntyre schrieb: You can use (n)awk in the case above. For example, what happens with cut in the case above if a future project allows usernames 8 characters? But granted that some people will have scripts that use cut, so it's hard to change. Compatible formatting if stdout is not a terminal. Optimized formatting if stdout is a terminal. Which would be really annoying to me as a user. I would go to look at the output, think I'll just pipe that to such and such command and then wonder why my parsing wasn't working... -- Less is only more where more is no good. --Frank Lloyd Wright Shawn Walker, Software and Systems Analyst [EMAIL PROTECTED] - http://binarycrusader.blogspot.com/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Shawn Walker wrote: On 19/05/07, Daniel Rock [EMAIL PROTECTED] wrote: Hugh McIntyre schrieb: You can use (n)awk in the case above. For example, what happens with cut in the case above if a future project allows usernames 8 characters? But granted that some people will have scripts that use cut, so it's hard to change. Compatible formatting if stdout is not a terminal. Optimized formatting if stdout is a terminal. Which would be really annoying to me as a user. I would go to look at the output, think I'll just pipe that to such and such command and then wonder why my parsing wasn't working... Ok, you have never tried piping the 'ls' command. Doug ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 19/05/07, Doug Scott [EMAIL PROTECTED] wrote: Shawn Walker wrote: On 19/05/07, Daniel Rock [EMAIL PROTECTED] wrote: Hugh McIntyre schrieb: You can use (n)awk in the case above. For example, what happens with cut in the case above if a future project allows usernames 8 characters? But granted that some people will have scripts that use cut, so it's hard to change. Compatible formatting if stdout is not a terminal. Optimized formatting if stdout is a terminal. Which would be really annoying to me as a user. I would go to look at the output, think I'll just pipe that to such and such command and then wonder why my parsing wasn't working... Ok, you have never tried piping the 'ls' command. My point is that it is unexpected behaviour. It doesn't matter if some other command does it or not. And no, I have never tried to parse the output of ls. I don't like magic switches -- and that is exactly what that behaviour feels like. A program that pretends to know better than me, and that's just infuriating. -- Less is only more where more is no good. --Frank Lloyd Wright Shawn Walker, Software and Systems Analyst [EMAIL PROTECTED] - http://binarycrusader.blogspot.com/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Shawn Walker schrieb: On 19/05/07, Daniel Rock [EMAIL PROTECTED] wrote: Compatible formatting if stdout is not a terminal. Optimized formatting if stdout is a terminal. Which would be really annoying to me as a user. I would go to look at the output, think I'll just pipe that to such and such command and then wonder why my parsing wasn't working... So how many scripts of yours failed in the past? % df -k [...] /usr/lib/libc/libc_hwcap2.so.1 12395698 7271171 500057160%/lib/libc.so.1 (output wrapped in two lines) % df -k | cat [...] /usr/lib/libc/libc_hwcap2.so.1 12395698 7271171 5000571 60% /lib/libc.so.1 (output in a single line) % ls /var adm cronldapnewsrun statmon apache dmi lib nfs sadmsvc [...] (multiple columns) % ls /var | cat adm apache apache2 appserver [...] (single column) etc. etc. You really shouldn't cut output by column number but by field number. And please don't forget to set LC_ALL=C in your scripts. I have seen lots of scripts which couldn't interpret the real size of verfügbar disk space while trying to parse a df -k output. Daniel ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 19/05/07, Daniel Rock [EMAIL PROTECTED] wrote: Shawn Walker schrieb: On 19/05/07, Daniel Rock [EMAIL PROTECTED] wrote: Compatible formatting if stdout is not a terminal. Optimized formatting if stdout is a terminal. Which would be really annoying to me as a user. I would go to look at the output, think I'll just pipe that to such and such command and then wonder why my parsing wasn't working... So how many scripts of yours failed in the past? As I said before, the point is that it is unexpected behaviour. Unless the man page for the utility explicitly lists the behaviour in question, it is undesireable in my view. Even then, I have misgivings about it. You also assume that I keep scripts around. I don't. I usually write one-liners in whatever shell I'm using when I have a need for something. -- Less is only more where more is no good. --Frank Lloyd Wright Shawn Walker, Software and Systems Analyst [EMAIL PROTECTED] - http://binarycrusader.blogspot.com/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Shawn Walker schrieb: As I said before, the point is that it is unexpected behaviour. Do you count the number of columns in an interactive context? Unless the man page for the utility explicitly lists the behaviour in question, it is undesireable in my view. Even then, I have misgivings about it. It is current practise in lots of utilities without a notice in a man page. So no difference to the sitiation today. You also assume that I keep scripts around. I don't. So what is your point? Daniel ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
The initial area of confusion hits with the distinction between packages and patches -- I know there's a difference between releasing functionality and fixing something that's broken. That's not a distinction, by and large, that is not made in the Linux world. If I'm running foo-1.1.1 and I need to update something, I find foo-1.1.2 or greater. There's no question that there's some downside with this approach, as new functionality can risk breakage, but that's what release notes are for. Similarly, package naming and dependency resolution go hand in hand. Why can't the package for foo-1.1.2 be named that instead of 118974-37, and why can't my attempt to add foo-1.1.2 at least notify me of the other packages I need to add to handle the dependencies and offer to get them for me.brbrLast night I was applying several security patches to Solaris 10 11/06. Of the 8, 6 failed to install with no information or explanation other than a failed notice which scrolled away from me fairly quickly. I don't know if it was a dependency, a configuration, user error. Had I not been watching it, I wouldn't have known it failed at all. I think some of this starts out with the development model. New work all takes place on the next release after the production release (and I suspect a tiny bit happens for the one after that even). Bug fixes get backported if they're likely to be a significant problem, or based on customer demand. Other bug fixes probably come about when a bug is found or reported in a supported release that's not apparent in the release under development. Bug fixes probably involve for the most part the smallest set of changes possible (simplifying testing, perhaps), although the scope of a patch grows in later revisions as fixes for additional bugs in the same and closely related files get added. Very rarely does a patch add a new package, it just makes minor updates to an existing one. I think another factor might be the historical distribution model, mostly via CDs or DVDs. What a patch usually is, is a partial overwrite of one or more packages, (i.e. of _some_ but not all of the files in a package), possibly accompanied by additional scripts to do smarter things like merge changes into configuration files. Along with that, the patches applied to any given package(s) are recorded, etc. I suppose the existing SVR4 pkg mechanism, give or take additional metadata added to each packages pkginfo file, is itself capable of other models, such as one where one always replaced entire packages. However, this would not speed downloads (bigger) or updates (more files to update), nor would it produce as much simplification as you might think, since a patch would still have to exist in the sense of total replacements of all of the packages that had to be updated together, plus any scripts needed to do tricky stuff. And the patches give you one thing by default that wholesale package replacement does not: the option to back them out. It's been worse in the past, but right now, I suspect that well under 1% of patches that successfully install, are such that one might need to back them out. But it still does happen occasionally. Also, patches as they are now (if they were a bit more careful about spelling out _why_ a reboot might be needed) potentially involve less reboots since they're replacing less files than whole package replacement. And keep in mind that a lot of this stuff is done the way it is for the sake of the sort of customer that probably schedules _any_ down time or maintenance well in advance. A lot of this isn't unique to Solaris; I think a number of other (mostly non-Linux) commercially distributed OSs distinguish between a package and a patch in some sense or another. Some of the reasons adding a patch might fail and the associated messages are described on the patchadd man page. Back when it was a (slow) script rather than a compiled program, the code also listed the meaning of all the return codes from patchadd. Since I can't find the source for the binary patchadd on src.opensolaris.org, I suspect it hasn't been opened (yet?), so I don't know whether those have changed. So I don't necessarily have a problem with SVR4 packages; I think they can be with perhaps additional metadata made quite capable of doing whatever is needed. I wouldn't rule out a binary file/package database, of which a text dump/restore could be made. (parallel binary and text files did I think exist for a little while, but had problems such that I think they went away again; not sure why, maybe maintenance (getting it right) cost exceeded anticipated performance benefits) And I don't necessarily even have a problem with some form of patches as a concept that builds on the concept of packages (which is pretty much the case now). I _do_ have a problem with the implementation of patches and of patching tools, although I haven't had an occasion to use the latest patching tools,
[osol-discuss] Re: Sun to make Solaris more Linux like
Sorry about that mess...here's another try. The initial area of confusion hits with the distinction between packages and patches -- I know there's a difference between releasing functionality and fixing something that's broken. That's not a distinction, by and large, that is not made in the Linux world. If I'm running foo-1.1.1 and I need to update something, I find foo-1.1.2 or greater. There's no question that there's some downside with this approach, as new functionality can risk breakage, but that's what release notes are for. Similarly, package naming and dependency resolution go hand in hand. Why can't the package for foo-1.1.2 be named that instead of 118974-37, and why can't my attempt to add foo-1.1.2 at least notify me of the other packages I need to add to handle the dependencies and offer to get them for me. Last night I was applying several security patches to Solaris 10 11/06. Of the 8, 6 failed to install with no information or explanation other than a failed notice which scrolled away from me fairly quickly. I don't know if it was a dependency, a configuration, user error. Had I not been watching it, I wouldn't have known it failed at all. I think some of this starts out with the development model. New work all takes place on the next release after the production release (and I suspect a tiny bit happens for the one after that even). Bug fixes get backported if they're likely to be a significant problem, or based on customer demand. Other bug fixes probably come about when a bug is found or reported in a supported release that's not apparent in the release under development. Bug fixes probably involve for the most part the smallest set of changes possible (simplifying testing, perhaps), although the scope of a patch grows in later revisions as fixes for additional bugs in the same and closely related files get added. Very rarely does a patch add a new package, it just makes minor updates to an existing one. I think another factor might be the historical distribution model, mostly via CDs or DVDs. What a patch usually is, is a partial overwrite of one or more packages, (i.e. of _some_ but not all of the files in a package), possibly accompanied by additional scripts to do smarter things like merge changes into configuration files. Along with that, the patches applied to any given package(s) are recorded, etc. I suppose the existing SVR4 pkg mechanism, give or take additional metadata added to each packages pkginfo file, is itself capable of other models, such as one where one always replaced entire packages. However, this would not speed downloads (bigger) or updates (more files to update), nor would it produce as much simplification as you might think, since a patch would still have to exist in the sense of total replacements of all of the packages that had to be updated together, plus any scripts needed to do tricky stuff. And the patches give you one thing by default that wholesale package replacement does not: the option to back them out. It's been worse in the past, but right now, I suspect that well under 1% of patches that successfully install, are such that one might need to back them out. But it still does happen occasionally. Also, patches as they are now (if they were a bit more careful about spelling out _why_ a reboot might be needed) potentially involve less reboots since they're replacing less files than whole package replacement. And keep in mind that a lot of this stuff is done the way it is for the sake of the sort of customer that probably schedules _any_ down time or maintenance well in advance. A lot of this isn't unique to Solaris; I think a number of other (mostly non-Linux) commercially distributed OSs distinguish between a package and a patch in some sense or another. Some of the reasons adding a patch might fail and the associated messages are described on the patchadd man page. Back when it was a (slow) script rather than a compiled program, the code also listed the meaning of all the return codes from patchadd. Since I can't find the source for the binary patchadd on src.opensolaris.org, I suspect it hasn't been opened (yet?), so I don't know whether those have changed. So I don't necessarily have a problem with SVR4 packages; I think they can be with perhaps additional metadata made quite capable of doing whatever is needed. I wouldn't rule out changes outside of the packages themselves. And I don't necessarily even have a problem with some form of patches as a concept that builds on the concept of packages (which is pretty much the case now). I _do_ have a problem with the implementation of patches and of patching tools, although I haven't had an occasion to use the latest patching tools, so I don't know if they don't stink as badly as the old approach of reading all the READMEs to determine the dependency order manually, and running patchadd (redirecting stdout, stderr to a file, and if you're smart, running lockfs -fa
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
--- Darren J Moffat [EMAIL PROTECTED] wrote: Chung Hang Christopher Chan wrote: And I don't see th package tools determining distribution models. Blastwave have a different distribution model from Sun and they use standard Solaris packages just fine. Can you store dependency data in Solaris packages? Yes, you have been able to store dependency data in Solaris packages since before rpm even existed. Okay, so extracting dependencies from solaris packages should not be a problem. Interesting. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
UNIX admin [EMAIL PROTECTED] wrote: HP-UX? Now that's a joke. hp being the braindead company they are, first killed off DECUnix (pardon, Tru64), and they haven't really done much of anything other than some minimal catching up, all while grinding their teeth, on HP-UX. I should know, since I have pretty recent HP-UX systems - at home! Do you have a HP-UX-11.x system? So, who exactly is left, other than Solaris, in the UNIX arena? There is nobody to set any standards, and there is nobody to follow them - just Solaris left not to break the existing ones. Any new standards will be set by Solaris for everybody else to follow. That's about the only UNIX left that is still innovating. Well, there is AIX but I am not sure whether IBM takes it for real and I know of no hacker who is using AIX as development platform. Making Solaris by default Linux alike looks really strange as most of the Linux innovations are just taken from other platforms (mainly Solaris). SunOS had the biggest influence on UNIX in the past 20+ years. We will not bring Solaris where it has been until 1995 by empty actions but by creating and showing new visions on the future of UNIX. Note that you will not get to the masses by vivions and note that the masses need only few things to be happy. Look at Mac OS X it is completely outdated from a OS point of view but people love it. In order to get to the massed you need to do few things to the immovable Linux users that allows them to easily _find_ their shell setup variant. But you will not really get new users this way. To get really new users that will bring you new customers in the future, you need to show new visions and bring it to the universities. Sun is very happy with the easy success in east Europe but conentrating to east Europe only is the wrong way. We need agressive approaches to universities in Western Europe and the USA. If you loose the market in the places that others still takes an example from you will again loose the complete market. Compared to other marketing activities from Sun, this would be cheap and the current idea of project Indiana looks to me like a Sun OpenSolaris distribution that (if done the way it currently seems) will most likely embrace and crush the sensitive plants that are the real free grown distributions. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Chung Hang Christopher Chan wrote: And I don't see th package tools determining distribution models. Blastwave have a different distribution model from Sun and they use standard Solaris packages just fine. Can you store dependency data in Solaris packages? Yes, you have been able to store dependency data in Solaris packages since before rpm even existed. -- Darren J Moffat ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 5/18/07, Chung Hang Christopher Chan [EMAIL PROTECTED] wrote: The whole development model is because there is no other way to do it since the packaging tools will not allow anything else. There is no choice but to create the patch system. Likewise the distribution model. That's simply not true. The way patches use packages is an implementation artefact - it says nothing about the underlying package technology. You could presumably build Sun-style patches using other packaging systems underneath (although not all packaging systems handle partial packages); you could build other patching or update systems on top of the current packaging system if you wanted to. Some update strategies may wish to enhance the underlying tools to make their own lives easier. That's fine - the pkg tools have been open source for over a year now, so you can enhance them if you wish. And I don't see th package tools determining distribution models. Blastwave have a different distribution model from Sun and they use standard Solaris packages just fine. -- -Peter Tribble http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
--- Richard L. Hamilton [EMAIL PROTECTED] wrote: Sorry about that mess...here's another try. The initial area of confusion hits with the distinction between packages and patches -- I know there's a difference between releasing functionality and fixing something that's broken. That's not a distinction, by and large, that is not made in the Linux world. If I'm running foo-1.1.1 and I need to update something, I find foo-1.1.2 or greater. There's no question that there's some downside with this approach, as new functionality can risk breakage, but that's what release notes are for. Similarly, package naming and dependency resolution go hand in hand. Why can't the package for foo-1.1.2 be named that instead of 118974-37, and why can't my attempt to add foo-1.1.2 at least notify me of the other packages I need to add to handle the dependencies and offer to get them for me. Last night I was applying several security patches to Solaris 10 11/06. Of the 8, 6 failed to install with no information or explanation other than a failed notice which scrolled away from me fairly quickly. I don't know if it was a dependency, a configuration, user error. Had I not been watching it, I wouldn't have known it failed at all. I think some of this starts out with the development model. New work all takes place on the next release after the production release (and I suspect a tiny bit happens for the one after that even). Bug fixes get backported if they're likely to be a significant problem, or based on customer demand. Other bug fixes probably come about when a bug is found or reported in a supported release that's not apparent in the release under development. Bug fixes probably involve for the most part the smallest set of changes possible (simplifying testing, perhaps), although the scope of a patch grows in later revisions as fixes for additional bugs in the same and closely related files get added. Very rarely does a patch add a new package, it just makes minor updates to an existing one. I think another factor might be the historical distribution model, mostly via CDs or DVDs. The whole development model is because there is no other way to do it since the packaging tools will not allow anything else. There is no choice but to create the patch system. Likewise the distribution model. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
And the patches give you one thing by default that wholesale package replacement does not: the option to back them out. You can also roll back on a package system. A lot of this isn't unique to Solaris; I think a number of other (mostly non-Linux) commercially distributed OSs distinguish between a package and a patch in some sense or another. Example please. I know of none. patches are not released. Updated packages are. And mapping patches to a whole package update approach is also tricky. Let's say that you have two patches that update a package, independent of another. You now have four possible versions of the package: original, patch A applied, patch B applied, or both. Having a repository have all those is unreasonable. Having it only have the most up to date for the release may sometimes not satisfy all possible dependency situations. It could be done, but it couldn't be done anything like automatically, IMO. I feel really reassured about a patch A + patch B updated package may not satisfy all possible dependency situations so I will accept both patches via the patch system. There are however cases where new packages get introduced for updates within a release. And there is an update mechanism for that, although it's IMO slow, ugly, hard to test, and often not as clean as a fresh install might be. The Live Upgrade approach (more or less: automatically building a replacement root/usr/var/whatever on an otherwise unused filesystem that becomes the new boot filesystem once it succeeds, and migrating config files as appropriate) is at least a lot safer (can always fall back) and can run with nothing but a performance impact while the system is in production. But there are times when it turns out to be perfectly ok to (starting with release X mm/yy) take packages from the next update, along with any new or updated dependencies, and add/update them manually onto the existing update. There are probably more such cases than are commonly recognized. Heck, I can think of more than one case where patch README files said to get a package from at least such and such an update and add it; and I _know_ there are more cases that sort of thing is never explicitly mentioned, but works out just fine. So I think that for any given release, it might perhaps not be impossible to have a repository based approach, but it would have to be supplemented by additional information _outside_of_ individual packages that caused packages to be aggregated as needed, additional scripts to be associated with those aggregations as needed, the process of applying a set of updates to (if applied to a running system rather than to an alternate boot environment as Live Upgrade does) be capable of being paused at various points, continued after some point in it that requires a reboot, restarted automatically in event of certain failures, etc. That is really inconvenient, having to keep your eye on two different software management systems. In effect, a repository based approach would have to provide all the benefits that the present version[/update] plus patches approach provides, be every bit as reliable, and a lot easier and more transparent. I certainly believe it does. Your example of a patch A + patch B versus updated package leads me to wonder how the solaris tools maintain dependencies... To merely do as some Linux distro does in terms of update management may well be far short of sufficient; the expectations of the established installation base are vastly different (stability and reliability; nice if it were easier, but it's always sucked anyway, so at least that's job security (back in SunOS 4.x (BSD based), patch installation was 100% manual, and sometimes even mildly technical (knowing C helped get a kernel patch installation right). So however bad the patch mechanism is now, what was before it was much worse, in fact nothing at all - not that that's an excuse!)). If you are referring to the srv4 package tools, then yes. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
And I don't see th package tools determining distribution models. Blastwave have a different distribution model from Sun and they use standard Solaris packages just fine. Can you store dependency data in Solaris packages? Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
On Fri, 18 May 2007, Jim Grisanzio wrote: Alan DuBoff wrote: OpenSolaris should belong to the community, the community should decide it's destiny, the community should be the sum of the entire community. Yes, Sun is a part of that, but it is only a part. It's not up to Sun to determine the destiny, IMO, and once revenue streams are attached to OpenSolaris, there seems more chance of that happening. Bottom line is that Sun needs to be careful with it's power, because it's not the only entity in the community. I think the Charter and Governance articulate pretty well that the OpenSolaris Community controls its own destiny: ARTICLE II: ... The OpenSolaris Community has the authority and responsibility for all decisions pertaining to the OpenSolaris software and collaborative infrastructure within the scope defined by the OpenSolaris Charter. http://www.opensolaris.org/os/community/ogb/governance / http://www.opensolaris.org/os/community/cab/charter.pd f Sure, but the problem with that is that the majority of the community and OGB is Sun employees. Sun can use their power to pay people to work on OpenSolaris, which is good, but it can also swing the scales in their favor. Sun is actually a massive a minority in the community at this point. There are 51,179 people registered on the site right now, and only about 2,000 of those have Sun badges. And that delta will only grow as we visit places like India and China, where the number of people who could potentially get involved in OpenSolaris is nothing short of stunning. Sure, the OGB is primarily Sun people, but that's who we elected. I doubt that situation will be the same in the future, but ultimately, it's up to community members to get involved and participate. Let's say the community does want to market OpenSolaris (I think that's kind of a dumb idea, but let's play along)...would they consult with Sun marketing do you think? I mean, has Sun marketing be that stellar in the past that those would be the folks that this community would turn to?gdr We have a Marketing Community -- and that community is as old as the project itself -- so it's up to the leaders in that community to assert themselves on marketing issues as they relate to community. I like the notion of marketing contributing to community development -- as I've said all along -- and they've done a very, very good job for us already. But I think we all confuse marketing for community development purposes and marketing for corporate product and competitive purposes. OpenSolaris should be about open development (engineering) and community building (like what you are doing with the SVOSUG), not about products and services and competitive issues with other companies and communities. I don't know if hiring Ian was good or not, but it was certainly better than some of the ads I've seen Sun run in the past, or the conent that has been on the website. Some of the posters I've seen at Sun are some of the worst marketing I've seen, ever. Stuff that is complicated, has so much info that it can't be absorbed, and leaves the viewer wondering, WTF is the point? while scratching their head. Now contrast some of that with some of Apples ads. A sillouette and an ipod, the only thing the sign says is ipod, and it just clicks with people. A large billboard of the MacBook Pro 17 and all it says on the TFT is Actual Size. These are effective ads. I don't think we should compare ourselves to Apple, and I don't think we should to compare ourselves to Linux, either. It's all a big distraction. I wonder if the Apple guys and the Linux guys talk about us as much as we talk about them. I doubt it. Take the other side of Sun Marketing, there was an ad with a single V880 in a lab. What is that about?:-/ With Apple ads you know what it's about somehow, there is no secrets. Apple ... no secrets? Humm ... :) Jim -- Jim Grisanzio, Sr. Program Manager, OpenSolaris Engineering http://blogs.sun.com/jimgris This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Chung Hang Christopher Chan [EMAIL PROTECTED] wrote: And I don't see th package tools determining distribution models. Blastwave have a different distribution model from Sun and they use standard Solaris packages just fine. Can you store dependency data in Solaris packages? This has been done since the early beginning. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 18/05/07, Chung Hang Christopher Chan [EMAIL PROTECTED] wrote: --- Richard L. Hamilton [EMAIL PROTECTED] wrote: Sorry about that mess...here's another try. The initial area of confusion hits with the distinction between packages and patches -- I know there's a difference between releasing functionality and fixing something that's broken. That's not a distinction, by and large, that is not made in the Linux world. If I'm running foo-1.1.1 and I need to update something, I find foo-1.1.2 or greater. There's no question that there's some downside with this approach, as new functionality can risk breakage, but that's what release notes are for. Similarly, package naming and dependency resolution go hand in hand. Why can't the package for foo-1.1.2 be named that instead of 118974-37, and why can't my attempt to add foo-1.1.2 at least notify me of the other packages I need to add to handle the dependencies and offer to get them for me. Last night I was applying several security patches to Solaris 10 11/06. Of the 8, 6 failed to install with no information or explanation other than a failed notice which scrolled away from me fairly quickly. I don't know if it was a dependency, a configuration, user error. Had I not been watching it, I wouldn't have known it failed at all. I think some of this starts out with the development model. New work all takes place on the next release after the production release (and I suspect a tiny bit happens for the one after that even). Bug fixes get backported if they're likely to be a significant problem, or based on customer demand. Other bug fixes probably come about when a bug is found or reported in a supported release that's not apparent in the release under development. Bug fixes probably involve for the most part the smallest set of changes possible (simplifying testing, perhaps), although the scope of a patch grows in later revisions as fixes for additional bugs in the same and closely related files get added. Very rarely does a patch add a new package, it just makes minor updates to an existing one. I think another factor might be the historical distribution model, mostly via CDs or DVDs. The whole development model is because there is no other way to do it since the packaging tools will not allow anything else. There is no choice but to create the patch system. Likewise the distribution model. I have not seen any proof of that, nor have you offered any. -- Less is only more where more is no good. --Frank Lloyd Wright Shawn Walker, Software and Systems Analyst [EMAIL PROTECTED] - http://binarycrusader.blogspot.com/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On Fri, 18 May 2007, Jim Grisanzio wrote: Sun is actually a massive a minority in the community at this point. There are 51,179 people registered on the site right now, and only about 2,000 of those have Sun badges. So, where are all those other people? I don't see them participating on the mailing lists. Those are marketing numbers, they mean nothing to what is done. This is no different than how Sun counts the downloads. Sure, the OGB is primarily Sun people, but that's who we elected. I doubt that situation will be the same in the future, but ultimately, it's up to community members to get involved and participate. Absolutely, and I agree it will change in the future. But until then the communities are dominated with Sun people and it's hard for anyone to have a pulse on each community without being subscribed. We need to understand that it exists how it is today. It is a fact that Sun has the largest influence with OpenSolaris as it is today. They will also continue to have a large influence for the future as well. From an engineering standpoint, this is good. From other areas of Sun, I don't know. OpenSolaris should be about open development (engineering) and community building (like what you are doing with the SVOSUG), not about products and services and competitive issues with other companies and communities. I agree with this statement, but let me explain something to you about SVOSUG and how I run it. There is very little input from Sun on what I should present, or how it should be presented. To date, I have looked at technologies that are happening in Solaris and lined up speakers to talk about them. I have not asked Sun for much in the way of funding, other than meeting space. Flip Russell provides drinks which are funded by Sun, but that is minimal. I have run many user groups, and been involved in many more. What I have learned is that these user groups are best grown out of grass roots, because it's the people who show up and are there to meet and talk with others that make up the group. Sure, we could get Sun to pay for food, pelt out more SWAG, and try to buy ourselves a user group. In the end that doesn't tend to work, IME, and would leave us with a Sun user group. I'm not saying we haven't had food or SWAG in the past, I just try to be selective, and not let it be overwhelming. Apple ... no secrets? Humm ... :) Secrets was not a good term to use, maybe hidden messages that are hard to figure out would have been better.;-) -- Alan DuBoff - Solaris x86 IHV/OEM Group ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On Thu, May 17, 2007 at 03:08:22PM +0700, Doug Scott wrote: The patch for df.c below checks to see if stdout is a tty (i.e. interactive), and will add the -h option if no other options conflict. If stdout is not a tty (i.e. a script), the df command is unchanged. Is there anything wrong with this idea?? isatty() has nothing to do with whether a command was run interactively. It simply says whether a file descriptor is attached to a terminal. So with your change typing df out or df | less wouldn't have the -h flag added. But running a script from the commandline with plain df in it would still have stdout attached to the terminal, and would have the -h flag added. Maybe there's something you can do if you're running from a modern shell that looks at process groups or sessions, but I'm not sure. Danek ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Andre van Eyssen wrote: It's a dose of hell for anyone learning to write scripts. I don't believe we need an envariable to enable this for several reasons: A) envariables don't scale - they are per-user, per-system, per-problem band-aids. Over time and over systems, this path leads to chaos. and, more importantly, B) we already have precedent in the system that this is an OK behavior: % uname -a SunOS sac 5.11 snv_56 sun4u sparc SUNW,Sun-Fire-V440 % /bin/ls Old sr.20070223.txt sr.20070406.txt sr.20070518.txt sr.20070119.txt sr.20070302.txt sr.20070413.txt sr.20070525.txt sr.20070126.txt sr.20070309.txt sr.20070420.txt sr.template.txt sr.20070202.txt sr.20070316.txt sr.20070427.txt sr.20070209.txt sr.20070323.txt sr.20070504.txt sr.20070219.txt sr.20070330.txt sr.20070511.txt % /bin/ls | cat - Old sr.20070119.txt sr.20070126.txt sr.20070202.txt sr.20070209.txt sr.20070219.txt sr.20070223.txt sr.20070302.txt sr.20070309.txt sr.20070316.txt sr.20070323.txt sr.20070330.txt sr.20070406.txt sr.20070413.txt sr.20070420.txt sr.20070427.txt sr.20070504.txt sr.20070511.txt sr.20070518.txt sr.20070525.txt sr.template.txt % -John ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
John Plocher [EMAIL PROTECTED] wrote: Andre van Eyssen wrote: It's a dose of hell for anyone learning to write scripts. I don't believe we need an envariable to enable this for several reasons: A) envariables don't scale - they are per-user, per-system, per-problem band-aids. Over time and over systems, this path leads to chaos. and, more importantly, B) we already have precedent in the system that this is an OK behavior: % uname -a SunOS sac 5.11 snv_56 sun4u sparc SUNW,Sun-Fire-V440 % /bin/ls Old sr.20070223.txt sr.20070406.txt sr.20070518.txt sr.20070119.txt sr.20070302.txt sr.20070413.txt sr.20070525.txt sr.20070126.txt sr.20070309.txt sr.20070420.txt sr.template.txt sr.20070202.txt sr.20070316.txt sr.20070427.txt sr.20070209.txt sr.20070323.txt sr.20070504.txt sr.20070219.txt sr.20070330.txt sr.20070511.txt % /bin/ls | cat - Old sr.20070119.txt sr.20070126.txt sr.20070202.txt This behavior is well known and well documented. Thos os different from new ideas. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Doug Scott schrieb: The patch for df.c below checks to see if stdout is a tty (i.e. interactive), and will add the -h option if no other options conflict. If stdout is not a tty (i.e. a script), the df command is unchanged. Is there anything wrong with this idea?? Why does everyone like the -h output? I personally hate it - and would hate it even more if it would be the default. With the -h option the width of the filesystem size is roughly the same - regardless if the filesystem is 1kB or 1TB. So I have to actually *read* the output instead of a quick look how wide the size column is. In order to compensate for larger disks and larger filesystem sizes I'd would much more like to see a -m flag (for the output in MBytes instead of kBytes). With then only 7 digits (instead of 10) even a filesystem of ~1TB could be easily checked. Daniel ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Danek Duvall wrote: On Thu, May 17, 2007 at 03:08:22PM +0700, Doug Scott wrote: The patch for df.c below checks to see if stdout is a tty (i.e. interactive), and will add the -h option if no other options conflict. If stdout is not a tty (i.e. a script), the df command is unchanged. Is there anything wrong with this idea?? isatty() has nothing to do with whether a command was run interactively. It simply says whether a file descriptor is attached to a terminal. So with your change typing df out or df | less wouldn't have the -h flag added. But running a script from the commandline with plain df in it would still have stdout attached to the terminal, and would have the -h flag added. Maybe there's something you can do if you're running from a modern shell that looks at process groups or sessions, but I'm not sure. Danek Ok, slightly bad wording by me :) I probably should have specified that the output is piped into another command to be further processed rather than be displayed. isatty(fileno(stdout)) can be used to give the desired result people have been asking for. i.e. Either for it to be read on the screen or processed by another command. I will remove the words interactive and scripts next time :) Doug ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
One thing to think about is how standards have changed.. It's no longer big vendors in a room deciding what the standard is (i.e., the top down approach). It's more the developers (largely in open source projects) deciding what the standard is as a side effect of writing their code.. How do we adapt to the new reality? Until an open collaboration of developers achieves the same documented process as industry or formal membership based bodies (ECMA, IEEE) or national (ANSI) or international standards bodies (ISO), _and_ there is a commitment to developing their standards in such a way that they are more or less platform neutral (which ultimately requires multi-platform development, i.e. participation of competing interests), I don't see how a group of developers can constitute any sort of a standard; at most, the product of their work will become the winner of an informal popularity contest. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Two things: Improvement can only take place with change or supplementation. If something does not improve, it will be replaced by superior alternatives. Sun wants Solaris to be successful, so change (or supplementation) must occur. If POSIX told you to hang yourself, you wouldn't do it, right? :) That said, I think the good news for the old guard is that the largest faults in Solaris can be fixed by supplementation rather than change. I'm referring to the GUI. Because whether the old guard knows it or not, what ls does today won't matter in the future because the new guard won't even be using it. (the crowd is shocked, but the world is not :)) On a set-top box or kiosk, all-GUI all the time is fine. On an appliance, I think one still might want access to more for greater diagnostics. On a general purpose system, a GUI becomes an oxymoron (since a menu cannot be general purpose because it consists of a limited, predefined list), and the user that limits themselves to only the GUI, unless they are indeed just interested in a special purpose system customized to their needs, is a moron. (excepting of course a _programmable_ (not merely minimally customizable) GUI environment, maybe something like the old Apple Hypercard) I am of course distinguishing between that class of users that does the same thing every day or over the course of every pay cycle or whatever (the informational equivalent of a production line), and that class of users that has no idea what they're doing the next day. The former depend on familiarity. The latter read the documentation. So if your new guard are mostly just informational production line types, I don't see how they'll be contributing to the advancement of anything, except insofar as large numbers of unsophisticated passive consumers do (sheep ripe for shearing, or who is stupid enough to pay for my product?). This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
So have a new environment variable SUNW_INCOMPATIBLE_LINUX_ACCOMODATION_DEFAULTS=true which if set causes various programs to change their defaults. Note that the programs would still have to add the functionality, just that the default behavior without that would remain as it is. That way, the newcomers could have their defaults, and I could have mine. Since they're the newcomers, they're the ones that have to do something extra, although there's no reason why that couldn't be something done just once (environment variable in per-account or per-system file) rather than every time they type a command, and there's no reason that the opportunities to set up such customizations couldn't go clear back to configurable defaults for useradd and even an extra option or two at installation time. So there would be no excuse for a claim of hardship on their part, IMO. That there _are_ ways to do this sort of thing is precisely why I think a separate new generation friendly distro is probably _not_ a good idea. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Another crazy idea would be to make utilities sensitive to a certain environment variable that defines the default flavor. It doesn't work out of the box exactly, but it could be made almost OOB if the choice of flavor is offered when the user account is created - and for most newbies that would also be install time, since the new installer will create the default non-root account - and automatically add that environment variable to ~/.profile. Nice idea, but it has one drawback: it still doesn't fix the scripting issues. You want consistent behavior in existing scripts, and the option of new behavior interactively and in new scripts that ask for it somehow. It seems to me that you need something added to all shells that allows for different behavior interactively vs in scripts. That could be aliases (if the shell should do the work) or it could be another environment variable (INTERACTIVE_INVOCATION=true) that would have to be set together with the UNIVERSE (or whatever) variable, and that the shell would explicitly not pass to a command run from a script (if the decision how to behave should be left to the command, given that it was provided sufficient information to know when altered defaults might be relatively harmless). I want to see something that only needs a single switch to set, per new account or per system, and that can be added to existing accounts by simply adding one line to the top of their .cshrc/.kshrc/.bashrc file (sh users are tougher, although one might fake aliases with functions up to a point). Beyond that aspect of simple setup and administration, and that stuff that currently runs on Solaris will be unaffected, and that it's maintainable and doesn't create new barriers to porting, I don't know that I care to much how it's done. But I really would rather see that integrated than see a separate distro as the means of accomplishing that sort of accomodation. With lots of systems, I'd just as soon have _one_ Solaris (at least in terms of a Sun distro, and not counting development releases like SXCE/SXDE), with simple per-system and per-account switches to tailor behavior; anything else would increase my maintenance load (subset of users want universe L behavior so I have to reload their workstation or do substantial customizations to alter its behavior or their account's? yuck). This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Compare what's happening around OpenSolaris distros to most Linux distros. Gentoo is amazing for docs and self-help; Ubuntu, Fedora, Debian and others are not that far behind. I have a rack of machines behind me, some Gentoo, some Ubuntu (all likely to be Debian soon). Diskless boxes, file servers, etc, and I was able to get them running almost entirely with Web-based resources (and some help from this list). Even stuff like RAID and ethernet bonding are well documented in how-to's, kernel documentation, etc. Likewise, the FreeBSD Handbooks and related docs are terrific and have been for a long time. OpenSolaris? No idea where to start. Putting up a file server on the right hardware, using NFS and ZFS? Totally no idea of where to start ;) This raises an interesting point. As a lot of comments regarding the lack of Linux documentation have been raised. The viewpoint of the Linux community is that OpenSolaris doesn't have enough documentation. Somehow we need to figure out what the disconnect is. I checked http://docs.opensolaris.org/ and http://www.opensolaris.org/docs but there wasn't anything there. I think we really need a newbie portal. (and link to it from everywhere) I think there's almost enough info out there between bigadmin and docs.sun.com and opensolaris.org and others, but I think it needs to be in a cohesive and navigable form, actually multiple such, so that one can have various views: introductory, transition from another platform, task oriented, reference, troubleshooting, and so on. I don't see those just as different ways of talking about the information (although there's certainly a difference in the actual text between task and reference descriptions, for example) but also as needing different maps with which to navigate the body of information. I also think it would be great if one could have the equivalent of an sdiff between related docs for different versions, supplemented by pointers to explicit discussion of transition/upgrade issues; eventually extended to handle not just release X vs Y, but to a lesser extent Linux vs Solaris specifics comparisons as well, where more or less parallel documentation already exists. It's not so much the lack of information, it's that publicity, navigation, access (docs.sun.com is _slow_ a lot of the time!), comprehensibility, and by comparison only occasionally completeness or correctness are lacking. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
IMHO, the Open Solaris community needs more than just programmers. Sure, but if someone that does documentation or marketing can code at least to the extent of the bite-size stuff, can in the former case read code without the need of constant consultation with the programmers, and in the latter case can comprehend documentation at least, then you don't have a bunch of disconnected functions all separately doing their things. They'd be capable of speaking about something to one another in common terms, even if when talking to consumers, they might use different language. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
--- Richard L. Hamilton [EMAIL PROTECTED] wrote: IMHO, the Open Solaris community needs more than just programmers. Sure, but if someone that does documentation or marketing can code at least to the extent of the bite-size stuff, can in the former case read code without the need of constant consultation with the programmers, and in the latter case can comprehend documentation at least, then you don't have a bunch of disconnected functions all separately doing their things. They'd be capable of speaking about something to one another in common terms, even if when talking to consumers, they might use different language. That's fine. That still does not contradict what I said about the Open Solaris community needing more than just programmers. Send instant messages to your online friends http://uk.messenger.yahoo.com ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Richard L. Hamilton wrote:
Another crazy idea would be to make utilities
sensitive to a certain
environment variable that defines the default
flavor. It doesn't work
out of the box exactly, but it could be made almost
OOB if the choice of
flavor is offered when the user account is created -
and for most
newbies that would also be install time, since the
new installer will
create the default non-root account - and
automatically add that
environment variable to ~/.profile.
Nice idea, but it has one drawback: it still doesn't
fix the scripting
issues.
You want consistent behavior in existing scripts, and the option of
new behavior interactively and in new scripts that ask for it somehow.
It seems to me that you need something added to all shells that
allows for different behavior interactively vs in scripts.
That could be aliases (if the shell should do the work) or it could be
another environment variable (INTERACTIVE_INVOCATION=true) that
would have to be set together with the UNIVERSE (or whatever) variable,
and that the shell would explicitly not pass to a command run from a script
(if the decision how to behave should be left to the command, given that
it was provided sufficient information to know when altered defaults might
be relatively harmless).
The patch for df.c below checks to see if stdout is a tty (i.e.
interactive), and will add
the -h option if no other options conflict. If stdout is not a tty (i.e.
a script), the df command
is unchanged. Is there anything wrong with this idea??
Doug
[EMAIL PROTECTED] df .
Filesystem size used avail capacity Mounted on
userdata/export/opensolaris/dev
92G 2.4G15G14%/export/opensolaris/dev
[EMAIL PROTECTED] df . | cat -
/export/opensolaris/dev(userdata/export/opensolaris/dev):31553909 blocks
31553909 files
[EMAIL PROTECTED] df -F ufs
Filesystem size used avail capacity Mounted on
/dev/dsk/c2d0s09.6G 7.3G 2.2G78%/
[EMAIL PROTECTED] df -F ufs | cat -
/ (/dev/dsk/c2d0s0 ): 4761064 blocks 944838 files
-
[EMAIL PROTECTED] hg diff usr/src/cmd/fs.d/df.c
diff -r f47ca9f35c1d usr/src/cmd/fs.d/df.c
--- a/usr/src/cmd/fs.d/df.c Tue May 01 05:33:55 2007 -0700
+++ b/usr/src/cmd/fs.d/df.c Thu May 17 14:55:13 2007 +0700
@@ -585,6 +585,17 @@ parse_options(int argc, char *argv[])
{
int arg;
+ /* if the stdout is connected to a terminal and the arguments do */
+ /* not conflict, the effectively add the -h option. */
+ int add_h_option;
+
+ if (isatty(fileno(stdout))) {
+ add_h_option = 1;
+ } else {
+ add_h_option = 0;
+ }
+
+
opterr = 0; /* getopt shouldn't complain about unknown
options */
#ifdef XPG4
@@ -602,16 +613,20 @@ parse_options(int argc, char *argv[])
V_option = TRUE;
} else if (arg == 'v' ! v_option) {
v_option = TRUE;
+ add_h_option = 0;
#ifdef XPG4
} else if (arg == 'P' ! P_option) {
SET_OPTION(P);
+ add_h_option = 0;
#endif
} else if (arg == 'a' ! a_option) {
SET_OPTION(a);
} else if (arg == 'b' ! b_option) {
SET_OPTION(b);
+ add_h_option = 0;
} else if (arg == 'e' ! e_option) {
SET_OPTION(e);
+ add_h_option = 0;
} else if (arg == 'g' ! g_option) {
SET_OPTION(g);
} else if (arg == 'h') {
@@ -619,10 +634,12 @@ parse_options(int argc, char *argv[])
scale = 1024;
} else if (arg == 'k' ! k_option) {
SET_OPTION(k);
+ add_h_option = 0;
} else if (arg == 'l' ! l_option) {
SET_OPTION(l);
} else if (arg == 'n' ! n_option) {
SET_OPTION(n);
+ add_h_option = 0;
} else if (arg == 't' ! t_option) {
SET_OPTION(t);
} else if (arg == 'o') {
@@ -631,11 +648,18 @@ parse_options(int argc, char *argv[])
the -o option can only be specified once);
o_option = TRUE;
o_option_arg = optarg;
+ add_h_option = 0;
} else if (arg == 'Z') {
SET_OPTION(Z);
} else if (arg == '?') {
errmsg(ERR_USAGE, unknown option: %c, optopt);
+ add_h_option = 0;
}
+
[osol-discuss] Re: Sun to make Solaris more Linux like
Yes, Sun does make a lot of the decisions today, in regard to OpenSolaris, and that is wrong, IMO. I believe the community should be responsible for doing their own management of their community and as we move forward that is happening more and more. It isn't said out loud too often, but there's got to be a limit to how much divergence between the community and Sun can practically exist with continuing results sufficient for the relationship to remain of value to both. I don't think that's a problem as long as neither falls into the mode of behavior that pushes the limits just to see how far they can go, or for reasons that are other than practical and openly stated. After all, both should benefit by improvements to and growing acceptance of [Open]Solaris. It might not be out of order for each to come up with an etiquette guide for their own. Some might hope that on Sun's side, that would include guidelines for community participation, avoiding the appearance of arrogance, and maximizing openness wherever possible. And some might hope that on the other side there was a bit more recognition that more of why things have been done the way they are is because it has _worked_, not just to pump up quarterly reports; in which case, differences shouldn't be about do/don't do this or you're wrong, they should be about _why_ is it done this way?. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
I chose to use words that are deliberately provocative, and engender some of the fears of agile methodology. Agile methods do emphasize real time communication, over written documents. Agile methods also emphasize working software as the measure of progress, and produce very little written documentation relative to other methods. This has resulted in criticism of agile methods as being undisciplined, which BTW is the point I was trying to emphasis with my choice of words. Where existing functionality for the user would not be broken, working is certainly better than nothing. But IMO, a distinction between working and production quality for high-reliability requirements would be essential, especially where the latter has historically been closer to the standard. I don't have one handy right now, but ISTR an Extras directory on Solaris install CDs, for stuff that was useful but not yet quite bundled, for whatever reason. Of course, for speed, there needs to be a package download approach rather than just an entire distribution on physical media. And for something still evolving rapidly, it seems to me all the more important to declare those aspects that are unstable, to discourage dependencies on unstable features or interfaces. Only through discipline can rapid and deep evolution coexist with acceptance of upgrades. As for documentation, IMO there had better at least be a commitment to develop it too, even if only in the form of a wiki or some such, so that as one converges on stability, a snapshot can be taken and the result rewritten into proper documentation. Heck, how can you do development at all with the possibility of non-zero turnover among the developers, unless there's something other than just the code to capture design choices and principles? So I think the more aggressive a model one chooses to follow, the more carefully one has to define the limits of the model. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Richard L. Hamilton wrote: I chose to use words that are deliberately provocative, and engender some of the fears of agile methodology. Agile methods do emphasize real time communication, over written documents. Agile methods also emphasize working software as the measure of progress, and produce very little written documentation relative to other methods. This has resulted in criticism of agile methods as being undisciplined, which BTW is the point I was trying to emphasis with my choice of words. Where existing functionality for the user would not be broken, working is certainly better than nothing. But IMO, a distinction between working and production quality for high-reliability requirements would be essential, especially where the latter has historically been closer to the standard. With agile (at least XP) working == production quality. As for documentation, IMO there had better at least be a commitment to develop it too, even if only in the form of a wiki or some such, so that as one converges on stability, a snapshot can be taken and the result rewritten into proper documentation. Heck, how can you do development at all with the possibility of non-zero turnover among the developers, unless there's something other than just the code to capture design choices and principles? A wiki is a very handy tool in those circumstances. So I think the more aggressive a model one chooses to follow, the more carefully one has to define the limits of the model. Whatever methodology is chosen, all aspects of the process have to be followed. All too often teams pretend to be following an agile process as an excuse for cutting corners. Ian ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On Thu, 17 May 2007, Doug Scott wrote: The patch for df.c below checks to see if stdout is a tty (i.e. interactive), and will add the -h option if no other options conflict. If stdout is not a tty (i.e. a script), the df command is unchanged. Is there anything wrong with this idea?? It's a dose of hell for anyone learning to write scripts. I would assume that this behaviour would be enabled by an environment variable, but there should be consistency between interactive and scripted output from simple commands. Imagine, for a minute, if that behaviour emerged on Linux. -- Andre van Eyssen. the only value you can add to a banana is a bruise -- McNealy. ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Andre van Eyssen wrote: On Thu, 17 May 2007, Doug Scott wrote: The patch for df.c below checks to see if stdout is a tty (i.e. interactive), and will add the -h option if no other options conflict. If stdout is not a tty (i.e. a script), the df command is unchanged. Is there anything wrong with this idea?? It's a dose of hell for anyone learning to write scripts. I would assume that this behaviour would be enabled by an environment variable, but there should be consistency between interactive and scripted output from simple commands. There are other examples such as the simple commands 'ls' and 'man' which do exactly this (do a grep of isatty in the cmd directory in the Solaris source). I would have thought that anybody learning scripts would actually do a man df, before they use the command (or if the output is not what the expected). I know I still do, many years later! Scripts will work as they have previously have and as documented. Doug ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Linux is the OS of GNU project, why can not SunOS be one of the OS of GNU project? Well, Hurd is actually the OS of the GNU project, but in practice Hurd never caught on or developed fast enough (and I've heard it has fundamental performance issues), so in practice, Linux is too, and the wording is perhaps better stated as an OS rather than the OS. And depending on the talk of dual CDDL/GPLv3-ing Solaris (and the possibility for that matter that GPLv3 may look a little more like CDDL in many ways than GPLv2 does), the notion of OpenSolaris having equivalent status in that regard to Linux is IMO interesting or at least amusing. In practice, there's Nexenta (OpenSolaris-based kernel/Debian based userland); there are also similar things with NetBSD and FreeBSD kernels. So for a distro, even the kernel license isn't necessarily a barrier. On the ideological side, I'm a bit more comfortable with ER than RMS... This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Coming from a Linux background, I don't expect OpenSolaris to look and feel like Linux, but I do expect the OpenSolaris community to actively engage with people like me. My impression is that, due to Sun's long proud history (and BSD roots), some in the OpenSolaris and Sun communities feel it is beneath them to recognize what Linux has been able to accomplish. This is not the way to win over Linux users, and I hope it is not the opinion of the majority of Solaris users. I hope so too. OTOH, some of the same goal and process differences that are responsible for some of Linux's relative strengths may also have played a part in those areas where it's relatively weak, and I think those already here might like to reduce barriers and adopt some of its strong points while preserving our own strengths (which include backwards compatibility). This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 5/16/07, Frank Van Der Linden [EMAIL PROTECTED] wrote: You can't do that with zones as they stand now. There's a whole list of things you can't do in a non-global zone. So you'd have to pick one environment for the global zone (which can do everything you want), and then one for another zone (which can't do a number of things). In other words, you'd be providing an environment for first-class citizens and an environment for second-class citizens, who will be stuck in a restricted environment. Zones are very cool for some usage models, like e.g. an ISP offering a restricted environment for clients. But there is a big mismatch between the model of selectable user environments and that of zones. I'm obviously working from the 50,000 foot view, but I thought you could run an entire copy of the OS (sans kernel) in a zone (as opposed to the sparse variety). The ideal situation would be if we could think of the global zone as the hypervisor, then make it easy to spin up virtual machines above that hypervisor (be they Xen domains or zones), with a few predefined environments people could choose from (GNU vs. Solaris classic). This goes a long way to pushing the unique advantages of Solaris angle that I keep talking about (ever tried to spin up a new Xen domain on Linux??) even as I suggest we need to borrow heavily from Linux. Plus, with ZFS, this makes the installation problem a lot easier too... P.S. - This approach helps with the laptop problem too, and again, in a very unique way: I just got a new laptop and am faced with the extremely painful task of migrating everything, which as anyone who has ever done this knows means move than just moving my home directory. This time, I installed Ubuntu under VMware, so next time, I'll just have to move the disk image. Instead of VMware, could the hypervisor be Solaris/OpenSolaris? -ian -- Ian Murdock 650-331-9324 http://ianmurdock.com/ Don't look back--something might be gaining on you. --Satchel Paige ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 5/17/07, Richard L. Hamilton [EMAIL PROTECTED] wrote: One thing to think about is how standards have changed.. It's no longer big vendors in a room deciding what the standard is (i.e., the top down approach). It's more the developers (largely in open source projects) deciding what the standard is as a side effect of writing their code.. How do we adapt to the new reality? Until an open collaboration of developers achieves the same documented process as industry or formal membership based bodies (ECMA, IEEE) or national (ANSI) or international standards bodies (ISO), _and_ there is a commitment to developing their standards in such a way that they are more or less platform neutral (which ultimately requires multi-platform development, i.e. participation of competing interests), I don't see how a group of developers can constitute any sort of a standard; at most, the product of their work will become the winner of an informal popularity contest. IETF? -ian -- Ian Murdock 650-331-9324 http://ianmurdock.com/ Don't look back--something might be gaining on you. --Satchel Paige ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Ian Murdock writes: On 5/17/07, Richard L. Hamilton [EMAIL PROTECTED] wrote: Until an open collaboration of developers achieves the same documented process as industry or formal membership based bodies (ECMA, IEEE) or national (ANSI) or international standards bodies (ISO), _and_ there is a commitment to developing their standards in such a way that they are more or less platform neutral (which ultimately requires multi-platform development, i.e. participation of competing interests), I don't see how a group of developers can constitute any sort of a standard; at most, the product of their work will become the winner of an informal popularity contest. IETF? There are some notable outlier cases, but the IETF primarily sets standards for the bits on the wire. In general, it's not at all concerned about user interface, operating system, or application design -- or even with coding -- which are exactly the issues we're discussing here. I don't see how that's relevant for the question at hand, which I believe was the importance of POSIX, SUS, and similar OS standards for modern systems (or the lack thereof). -- James Carlson, Solaris Networking [EMAIL PROTECTED] Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Casper writes: As I see it, there are several categories: - missing utilities (functionality, command line options, ...) - no brainer: just add them/it - differences in behaviour: - backspace/delete : unfortunate mistakes made with the introduction of the IBM PC are hard to rectify but the observed differences is mostly due to the fact that command line editing is not on by default - history up/down (see above) - command output differences (df vs df -k) - several solutions proposed - difference of the order awk/gawk - PATH? I don't think it goes much deeper than that. The reason I may prefer PATH over other solutions is that scripts when written correctly will set $PATH at the start and setting PATH will cause them to be correct. Environment variables or other magic execution environment settings all suffer from the same problem: it is difficult to conceive of a solution which will cause already running scripts to continue to work. Casper As a SunOS user who switched to Linux in 1993 and has since reintroduced myself to Solaris 9/10, I can tell you the primary difference for me is software and package management. The limited default shell is annoying, but is easily solvable. The differences in command output and switches don't bother me much either -- I expect different *nixes to be different in small ways. Software management, however, feels like a crippling lack to me. The initial area of confusion hits with the distinction between packages and patches -- I know there's a difference between releasing functionality and fixing something that's broken. That's not a distinction, by and large, that is not made in the Linux world. If I'm running foo-1.1.1 and I need to update something, I find foo-1.1.2 or greater. There's no question that there's some downside with this approach, as new functionality can risk breakage, but that's what release notes are for. Similarly, package naming and dependency resolution go hand in hand. Why can't the package for foo-1.1.2 be named that instead of 118974-37, and why can't my attempt to add foo-1.1.2 at least notify me of the other packages I need to add to handle the dependencies and offer to get them for me. Last night I was applying several security patches to Solaris 10 11/06. Of the 8, 6 failed to install with no information or explanation other than a failed notice which scrolled away from me fairly quickly. I don't know if it was a dependency, a configuration, user error. Had I not been watching it, I wouldn't have known it failed at all. I want access to dtrace and ZFS, I want zones and I want trusted Solaris extensions. That's why I'm returning after a lot of years. Presenting the Linux user with a more familiar software management mechanism will go a long way toward solving the familiarity issues. If I can get the software (and check what I have already), I can make my scripts or the interactive pieces of my environment behave like I want them to behave. I don't mind having to get (or start) a better shell, change PATH to give me a familiar ps or such. When it takes me 10 minutes and 2 google searches to get the package manager to tell me if and what version of postgres I have installed, however, I'll be turned off pretty quickly. It doesn't have to be apt or rpm. But moving from an rpm based distro to an apt/deb based distro took me 3 minutes and 2 man pages before I was comfortable. I know much of this has been discussed and is being addressed through projects like Nexenta. I don't need the whole Linux userland, however, to be comfortable with making a switch. With good repositories and package management, I'd be comfortable starting with a more traditional Solaris userland confident that I could get and maintain the pieces that just didn't work for me for whatever reason. Rich ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
The opposite could be said about Solaris. A) Solaris has an illustrious history of adopting useless standards, You've never truly lived until you've been dumped in a middle of a salad of all kinds of Linux distros - from RedHat 9 to SuSE 9 to RHES to RHEL to ... Then you'll know what pains and nightmares look like during waking hours. And you'll wish you could take what you wrote above back - about 1000 times over. and b) solaris has an illustrious history of not meeting a needs if there is any conflict with a standard. How about some facts? Which conflicting standards does or did Solaris implement? IMHO - Solaris is not Linux, nor should it ever be. This we agree on. We all do - the day Solaris becomes Linux is the day I quit working with computers. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
Solaris has many features which should be moving people from Linux to Solaris on mass. Unfortunately, a lot are turned off by some things which should have been fixed in Solaris a long time ago. Hopefully Project Indiana will address these. Just because Solaris doesn't function like Linux does not make it broken. (Fix implies something is broken.) What exactly do you believe to be broken in Solaris? A question. When is the next Unix standard expected to be finalised? Do you at least now the year, and who is going to release it? What is it going to be based on? Does it really matter? Except for Solaris, which UNIX out there is really innovating? SGI IRIX hit the way of the dodo when SGI in its infinite wisdom axed it and decided to beg and plead mercy before the sitting penguin. Like that's gonna save them! Meanwhile hardcore IRIX users are migrating en masse, not to Linux or HP-UX, but get this - to Solaris on SPARC! And judging by what they're writing on nekochan.net, they like Solaris. As well they ought to, since they migrated from one System V UNIX to another. HP-UX? Now that's a joke. hp being the braindead company they are, first killed off DECUnix (pardon, Tru64), and they haven't really done much of anything other than some minimal catching up, all while grinding their teeth, on HP-UX. I should know, since I have pretty recent HP-UX systems - at home! So, who exactly is left, other than Solaris, in the UNIX arena? There is nobody to set any standards, and there is nobody to follow them - just Solaris left not to break the existing ones. Any new standards will be set by Solaris for everybody else to follow. That's about the only UNIX left that is still innovating. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
So, who exactly is left, other than Solaris, in the UNIX arena? IBM?, AIX Micro, Dyanmic, and VIO Partitions are kinda neat and they are supported on their big database servers as opposed to Sun Logical Domains. Granted this is mostly a firmware/hardware hack, but hey we need some other aspects to discuss. Jordan On 5/17/07, UNIX admin [EMAIL PROTECTED] wrote: Solaris has many features which should be moving people from Linux to Solaris on mass. Unfortunately, a lot are turned off by some things which should have been fixed in Solaris a long time ago. Hopefully Project Indiana will address these. Just because Solaris doesn't function like Linux does not make it broken. (Fix implies something is broken.) What exactly do you believe to be broken in Solaris? A question. When is the next Unix standard expected to be finalised? Do you at least now the year, and who is going to release it? What is it going to be based on? Does it really matter? Except for Solaris, which UNIX out there is really innovating? SGI IRIX hit the way of the dodo when SGI in its infinite wisdom axed it and decided to beg and plead mercy before the sitting penguin. Like that's gonna save them! Meanwhile hardcore IRIX users are migrating en masse, not to Linux or HP-UX, but get this - to Solaris on SPARC! And judging by what they're writing on nekochan.net, they like Solaris. As well they ought to, since they migrated from one System V UNIX to another. HP-UX? Now that's a joke. hp being the braindead company they are, first killed off DECUnix (pardon, Tru64), and they haven't really done much of anything other than some minimal catching up, all while grinding their teeth, on HP-UX. I should know, since I have pretty recent HP-UX systems - at home! So, who exactly is left, other than Solaris, in the UNIX arena? There is nobody to set any standards, and there is nobody to follow them - just Solaris left not to break the existing ones. Any new standards will be set by Solaris for everybody else to follow. That's about the only UNIX left that is still innovating. This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
[osol-discuss] Re: Sun to make Solaris more Linux like
You know what, I totally disagree with this move: Don't make Solaris Linux like, BUT teach us Linux guys the Solaris way. As I read here again and again the POSIX way - what ever that means, at least I don't know, and I am sure many young(as in age and as in new to Unix) Linux users don't know,too -. http://www.unix.org/single_unix_specification/(requires sign-up) The Single Unix Specification is actually a superset of POSIX; the POSIX standard by itself is from IEEE and probably not freely available, but is effectively duplicated by US Government specifications and by ISO/IEC standards; for a body coordinating updates for POSIX and the core of the Single Unix Specification, see http://www.opengroup.org/austin/ This means that where the standards says undefined, it's best to consider any such feature or behavior (if there is any consistent behavior on any platform) as non-portable, and when it says implementation defined, that means there may be optional implementation specific behavior but again, if you want maximum portability, don't go there. At least as much of the discipline in writing code that's both portable and maintainable is in what one leaves out. That doesn't necessarily preclude friendliness or features, but it means that everything is done deliberately; by comparison some open source seems to consider it enough that it's open, and aspire to becoming a defacto rather than a documented standard. It also means that not everyone cares about source availability except insofar as it clearly connects with an improved experience and lower prices. This is about what's been demonstrated to work, not about philosophy. So the we don't need no steenkin' ABI* model just doesn't cut it, and backwards compatibility (not breaking existing binaries that confine themselves to documented-as-stable interfaces; and to some extent, that accomodating a new user base should not require either recoding or retraining by the established user base) is very important. It means that new features, whether as part of accomodating a changing user base or for any other reason, need to go through a consistent process, not just a couple of people convincing a single gatekeeper to accept them; in turn, that means lots of patience and arguments (which for years probably went on just as vigorously as any of the Linux mailing list flame fests, but until fairly recently were not a public spectacle). * http://www.darryl.com/badges/ This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Ian Murdock wrote: Well, if that's what you think I've said so far, I haven't done a very good job of articulating my thoughts. Bottom line is: We have Zones, so we can provide two environments, one for people who are perfectly happy with Solaris as it exists today, and another for people who are more familiar with the Linux environment. So, the fact that we don't agree which one is better isn't a showstopper. You can't do that with zones as they stand now. There's a whole list of things you can't do in a non-global zone. So you'd have to pick one environment for the global zone (which can do everything you want), and then one for another zone (which can't do a number of things). In other words, you'd be providing an environment for first-class citizens and an environment for second-class citizens, who will be stuck in a restricted environment. Zones are very cool for some usage models, like e.g. an ISP offering a restricted environment for clients. But there is a big mismatch between the model of selectable user environments and that of zones. - Frank ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Frank Van Der Linden wrote: Ian Murdock wrote: [...] perfectly happy with Solaris as it exists today, and another for people who are more familiar with the Linux environment. So, the fact that we don't agree which one is better isn't a showstopper. [...]environment for first-class citizens and an environment for second-class citizens, who will be stuck in a restricted environment. Zones are very cool for some usage models, like e.g. an ISP offering a restricted environment for clients. But there is a big mismatch between the model of selectable user environments and that of zones. Nexenta has a very nice approach to this problem. You can set an environment variable to control your userland environment personality. I believe they have hooks into exec to determine the personality of the caller and automatically resolve binaries from /usr/sun if SUN_PERSONALITY == 1. Avoids having to fiddle around with your PATH. It is per-process so you can execute a command that expects a Solaris compatible env just by setting this variable from a GNU env thusly: SUN_PERSONALITY=1 cmd. Regards, Moinak. - Frank ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Moinak Ghosh wrote: Nexenta has a very nice approach to this problem. You can set an environment variable to control your userland environment personality. I believe they have hooks into exec to determine the personality of the caller and automatically resolve binaries from /usr/sun if SUN_PERSONALITY == 1. Avoids having to fiddle around with your PATH. It is per-process so you can execute a command that expects a Solaris compatible env just by setting this variable from a GNU env thusly: SUN_PERSONALITY=1 cmd. That sounds like a better approach, although the problem with that kind of magic is that it's not transparent. I wonder, for example, how that would work with something like the internal executable path table that *csh has. Or with commands such as 'which' and 'type'. But, it's certainly something that looks like a better approach than zones, since those are like a screwdriver to this nail. - Frank ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Frank Van Der Linden wrote: Moinak Ghosh wrote: Nexenta has a very nice approach to this problem. You can set an environment variable to control your userland environment personality. I believe they have hooks into exec to determine the personality of the caller and automatically resolve binaries from /usr/sun if SUN_PERSONALITY == 1. Avoids having to fiddle around with your PATH. It is per-process so you can execute a command that expects a Solaris compatible env just by setting this variable from a GNU env thusly: SUN_PERSONALITY=1 cmd. That sounds like a better approach, although the problem with that kind of magic is that it's not transparent. I wonder, for example, how that would work with something like the internal executable path table that *csh has. Or with commands such as 'which' and 'type'. It is in fact completely transparent since the path translation happens in the kernel side of the the exec syscall. Regards, Moinak. But, it's certainly something that looks like a better approach than zones, since those are like a screwdriver to this nail. - Frank ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Moinak Ghosh wrote: Nexenta has a very nice approach to this problem. You can set an environment variable to control your userland environment personality. I believe they have hooks into exec to determine the personality of the caller and automatically resolve binaries from /usr/sun if SUN_PERSONALITY == 1. Avoids having to fiddle around with your PATH. It is per-process so you can execute a command that expects a Solaris compatible env just by setting this variable from a GNU env thusly: SUN_PERSONALITY=1 cmd. That sounds like a better approach, although the problem with that kind of magic is that it's not transparent. I wonder, for example, how that would work with something like the internal executable path table that *csh has. Or with commands such as 'which' and 'type'. Why would that not be transparent? (I'm assuming we're mapping paths in /usr/bin to /usr/sun *inside* the kernel) But using environment variables inside the kernel is a bit of an issue. But, it's certainly something that looks like a better approach than zones, since those are like a screwdriver to this nail. You'd still need to figure out a way to run old scripts transparently, regardless of environment settings. Casper ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
[EMAIL PROTECTED] wrote: Why would that not be transparent? (I'm assuming we're mapping paths in /usr/bin to /usr/sun *inside* the kernel) The csh hash table was a bad example, that one should be ok. However, it's not 100% transparent, since it looks to me like an application that would do: fd = open(/usr/bin/ls, O_RDONLY); .. do some checks on the file.. ..file is ok, let's exec it... execvp(/usr/bin/ls, ...) That application would be rather surprised that 's getting /usr/sun/ls (or whatever) instead. I'm not saying that this would be a regular occurence, but it's not quite transparent. Of course, at this point I should probably stop talking and see how they actually implemented it :-) - Frank ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Richard L. Hamilton [EMAIL PROTECTED] wrote: You know what, I totally disagree with this move: Don't make Solaris Linux like, BUT teach us Linux guys the Solaris way. As I read here again and again the POSIX way - what ever that means, at least I don't know, and I am sure many young(as in age and as in new to Unix) Linux users don't know,too -. http://www.unix.org/single_unix_specification/(requires sign-up) ... It also means that not everyone cares about source availability except insofar as it clearly connects with an improved experience and lower prices. This is about what's been demonstrated to work, not about philosophy. So the we don't need no steenkin' ABI* model just doesn't cut it, and backwards compatibility (not breaking existing binaries that confine themselves to Please note that POSIX (ot SUS) is a source standard. It does not define e.g. paths. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Moinak Ghosh wrote: Frank Van Der Linden wrote: Moinak Ghosh wrote: Nexenta has a very nice approach to this problem. You can set an environment variable to control your userland environment personality. I believe they have hooks into exec to determine the personality of the caller and automatically resolve binaries from /usr/sun if SUN_PERSONALITY == 1. Avoids having to fiddle around with your PATH. It is per-process so you can execute a command that expects a Solaris compatible env just by setting this variable from a GNU env thusly: SUN_PERSONALITY=1 cmd. That sounds like a better approach, although the problem with that kind of magic is that it's not transparent. I wonder, for example, how that would work with something like the internal executable path table that *csh has. Or with commands such as 'which' and 'type'. It is in fact completely transparent since the path translation happens in the kernel side of the the exec syscall. I have seen something similar used in the past to get around CPU instruction set differences. On Sony's (Yes Sony) version of BSD (possibly others), they were able to use an environment variable within a symbolic link. It actually worked really well. There have been many times since when I could have used it on other versions of Unix. Doug ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On Tue, May 15, 2007 at 06:46:39PM -0400, Ian Murdock wrote: On 5/13/07, Ceri Davies [EMAIL PROTECTED] wrote: On Sat, May 12, 2007 at 03:01:32PM -0700, MC wrote: Two things: Improvement can only take place with change or supplementation. If something does not improve, it will be replaced by superior alternatives. Sun wants Solaris to be successful, so change (or supplementation) must occur. If POSIX told you to hang yourself, you wouldn't do it, right? :) I think the major point for Ian to be be taking away from here is that if he thought he could turn up to Sun and change the default behaviour of Solaris because nothing would break probably and gee it up by putting a bunch of exclamation points in his emails to the community then he probably joined the wrong company. Sun has users that depend on the default behaviour of Solaris, they're the users who've spent all the money so far (because apparently nobody else gives Sun any money due to csh) and breaking their stuff, or making them even question whether their stuff is going to break, is a very bad idea indeed. Well, if that's what you think I've said so far, I haven't done a very good job of articulating my thoughts. Bottom line is: We have Zones, so we can provide two environments, one for people who are perfectly happy with Solaris as it exists today, and another for people who are more familiar with the Linux environment. So, the fact that we don't agree which one is better isn't a showstopper. Agreed. I don't see how it would work, but I am happy so long as backwards compatibility is retained by default. Ceri -- That must be wonderful! I don't understand it at all. -- Moliere pgpIyfUV5QFXP.pgp Description: PGP signature ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Doug Scott [EMAIL PROTECTED] wrote: I have seen something similar used in the past to get around CPU instruction set differences. On Sony's (Yes Sony) version of BSD (possibly others), they were able to use an environment variable within a symbolic link. It actually worked really well. There have been many times since when I could have used it on other versions of Unix. This was a feature of Apollo Domain OS. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED](uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
Doug Scott [EMAIL PROTECTED] wrote: I have seen something similar used in the past to get around CPU instruction set differences. On Sony's (Yes Sony) version of BSD (possibly others), they were able to use an environment variable within a symbolic link. It actually worked really well. There have been many times since when I could have used it on other versions of Unix. This was a feature of Apollo Domain OS. I seem to remember PrimeOS or some such had this too. (Or did that have the magic context sensitive links) Casper ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On Wed, 16 May 2007, Frank Van Der Linden wrote: You can't do that with zones as they stand now. There's a whole list of things you can't do in a non-global zone. That is true today, but Xen might change that. We do have branded zones today which run a Linux personality, but as you point out, non-global zones have some limitations, but I see that as a good thing. -- Alan DuBoff - Solaris x86 IHV/OEM Group ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] Re: Sun to make Solaris more Linux like
On 16/05/07, Alan DuBoff [EMAIL PROTECTED] wrote: On Wed, 16 May 2007, Frank Van Der Linden wrote: You can't do that with zones as they stand now. There's a whole list of things you can't do in a non-global zone. That is true today, but Xen might change that. We do have branded zones today which run a Linux personality, but as you point out, non-global zones have some limitations, but I see that as a good thing. This strikes me as massive overkill. Why use a zone (let alone a full VM) to avoid a user having to set their PATH correctly? -- Rasputin :: Jack of All Trades - Master of Nuns http://number9.hellooperator.net/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
