Re: [SLUG] Re: Why XML bites and why it is NOT a markup language
We're probably abusing the list here ... You can imprint a record-oriented structure onto a stream format by using tags in the stream but trying to support a stream by using a record format is really ugly (not impossible). It is desirable to have a format that makes it easy to build higher level formats on top of rather than a format which is already so high level that it becomes cumbersome for ordinary tasks. Going back to my DVB example, the MPEG data is a natural fit inside of 188 byte blocks. However, the program guides and other info aren't. So they have sequence numbers, continuation markers and other sundry transport stuff. Nasty, but neccessary because of temporal data needs. Providing you don't want to seek an XML stream... Seeking. Ohoh, you've introduced new material at a late stage! Seeking and stream oriented may be asking a bit much. Seeking is very hard when you have escape characters because the block lengths may not be exactly the same. Not at all. Validation is a higher level function and should be treated as such. Building a layered architecture is far more reliable, flexible and maintainable than building a monolithic architecture. Thus the job of the parser is to: read the raw data stream; identify the tags; identify the data blocks and provide an API that gives access to these entities (and nothing else). I guess xml proponents would be saying that xml is that higher layer. XML does not remove the requirement to sanity-check the data you are given. Yes, that is true. That is why I mentioned UBF in a previous email. It has a validation stage which is based on a contract definition between two applications. belongs to... but all that is an optional add-on. As I mentioned earlier, you can take a robust protocol and deliberately make it brittle if that is useful for your application but you can't go the other way. So start with something simple and robust and add the brittle bits if and when you happen to need them. This is the bit I don't buy. You seem to want to add redundancy to get around problems with a transport layer (or is it human mistakes?) If it's transport problems, then you need to design something special for the kind of transport you are supporting, e.g. like DVB does. If it's human problems, well, I just don't think you'll ever solve the problem of humans getting away with the least amount of effort... Then again, you could look at how Wikipedia keeps metadata and that also supports high quality resynchronisation because it depends on particular patterns being detected and anything that doesn't match a ... I mean, if XML is so fantastic, why does every Wiki avoid it (for the wiki data pages at any rate)? Now you're really getting to the meat of what many people find wrong with xml. It's butt-ugly. It requires a fairly large parser. No-one wants to type in raw xml. ... until you want to seek, or someone manages to introduce a single byte of bad data into your database though perhaps a bug or simple user idiocy that no programmer was expecting, or someone wanted to tweak a record in the data but there was no program function for it so they just tweaked the record in a text editor (and screwed something at the same time) or all of those other things that really do happen especially when everyone is sure they can't happen. Now hang on a minute. We've just gone down the road of special escape characters, low ascii characters, and maximum transport lengths. You can't use an editor successfully with these. 'Someone just tweaked a record'. See, you can't have unreliable data and a bunch of ad-hoc rules and end up with reliable data. Maybe the parser could give you more sensible warnings and errors, yes, but it still can't fix your mistakes reliably. I'm 100% confident that XML will cut itself off at the knees. Programming fads go in approximately 10 year cycles and I think XML is about 5 years in and already people are asking why bother? Well, xml is a simplified version of SGML, meant to strip out the over-complicated features of SGML. I have a copy of Goldfarb's SGML Handbook, published in 1990. In it, it is claimed the first working draft of the SGML standard was published in 1980. That's 25 years. I think the trend is towards more correctness of data, more strictness in specifying inter-process communication, away from ad-hoc approaches. There are problems with xml in certain domains; it's verbose, it's not very readable, it's expensive to parse for certain applications. It doesn't lend itself to every problem domain, and yet is often pushed to do so (soap, xlst to name a few off the top of my head). The other cool thing is that since XML is so nicely structured, any existing documents that DO go through XML parsers can rapidly be converted to any other tagged format. If a new format comes along that does everything XML does AND is easier to use AND more robust AND supports more documents then adoption
Re: [SLUG] Ubuntu kernel upgrade woes
On Fri, Jun 10, 2005 at 07:37:52 +1000, Michael Chesterton wrote: I don't know what happened in your situation, you might have been unlucky, or hit a bug. It usually just works. IIRC the kernel was upgraded when I did an 'apt-get upgrade' after the initial installation and it worked fine. found the bug https://bugzilla.ubuntu.com/show_bug.cgi?id=11135 Thanks. I did install a package from breezy (nedit-5.5 because the nedit-5.4 in hoary is broken) so it looks like that was my problem. I'm now trying again, this time building nedit-5.5 from source because the nedit-5.5 binary from nedit.com doesn't work on hoary (but it works fine on RH 7.3). Cheers, John -- Being NT servers, how about warez, mp3z, pr0n, vcdz, etc. This way you have named them after the likely rĂ´le they'll have once the script kiddies find them. -- Peter Corlett -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Ubuntu kernel source and binary packages
On Fri, Jun 10, 2005 at 07:51:35 +1000, Michael Chesterton wrote: I got confused by this, too. The package name has a version number in it, which is independent to the package Version: number. $ apt-cache policy linux-image-2.6.10-5-686-smp Thanks again. I'm slowly learning how apt/dpkg works. Cheers, John -- Someone should write a paper on the comparative efficacy of garlic and ramps in dealing with the undead. I find very strong coffee works much better on me in the morning. -- Peter Corlett -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
RE: [SLUG] Ubuntu kernel upgrade woes
The exact same thing happened to me. I normally update anything available. The next time I swiched on my laptop the day after it gave me the same unable to boot error. Had to install from scratch. Good thing I backed up my data a week before but yeah it irritated me big time. Carlo -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of John Clarke Sent: Friday, 10 June 2005 11:58 AM To: slug@slug.org.au Subject: [SLUG] Ubuntu kernel upgrade woes Hi all, Ubuntu released a new kernel yesterday, which I duly installed with 'apt-get upgrade'. I'm used to keeping the old kernel in place until I've verified that the new one boots (rpm allows two different versions to be installed at the same time), but I couldn't figure out how to do that with apt, so I just let it replace the old one. No errors were reported so I rebooted. I know, I should have just told it to upgrade the 686-smp kernel so that the 386 kernel was there in case anything went wrong, but I realised a few seconds too late that I'd managed to replace both known good kernels and had nothing to fall back on. You've probably guessed what happened next. Neither of the new kernels would boot -- the error was VFS: Unable to mount root fs on unknown block (0,0). I tried for about half an hour to fix it, but without success, so I reinstalled from scratch. It's not my favourite way to fix a problem, but I figured it'd be the quickest way to get the system back up. I hadn't done much since installing it last week all I've lost is time. Once the installation had finished, I installed the 686-smp kernel and rebooted. I expected it not too boot, but this time it came up fine. So, what did I do wrong during the upgrade (other than not keeping a known good kernel around)? Is there some additional magic required when upgrading kernels with 'apt-get upgrade'? Thanks, John -- COBOL is English like. That is, assuming that your English has first been gang banged by a herd of irate kudzu and is suffering from a digestive upset after consuming all the Hydrogen Selenide. -- Shmuel (Seymour J.) Metz -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html _ This message has been checked for all known viruses by the MessageLabs Virus Scanning Service. For further information visit http://www.Hi-Speed.net.au __ __ This email has been scanned for all viruses by the MessageLabs SkyScan service. For more information on a proactive anti-virus service working around the clock, around the globe, visit http://www.hi-speed.net.au -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
[SLUG] Seeking User Group Management Advice
Sluggers, I'm looking for some advice in regards to user and group account management. I have a piece of software (Matlab) for which we have a single user licence. I.e the licence is tied to a user account on the Linux server. In order to use this software you have to login as that user. Or login as yourself and su - matlab. What is going to be the most elegant method for allowing users to run this software (one at a time) and save their results back to their home directory without being able to trash each others account? Is there an elegant way to become another user but retain your group privledge? Perhaps something like login as peter then su - matlab ; newgrp peter? When I do this it prompts for a group password, where is this kept and how do I set it? (man newgrp tells me about the passwd but not how to set it). Ideally this would all be done in a script so the user clicks on the matlab icon on their icewm desktop (served via ltsp or vnc) and then they key in a passwd or two and volia matlab runs but they retain the ability to open/save work in their home directory. Does anyone know of an open source matlab alternate? The Linux server will provide, *nix logins, Imap e-mail, possibly some protected web pages with Apache, and samba shares of their home directory to their desktop PC. I've been considering using LDAP but on past experience found it to be quite a lot to chew, there will only be about 40 users.Any opinions on whether LDAP is worth the effort. I would need users to be able to set their passwords and have that ripple through to each service. Currently PC's are just in a work group, trying to decide whether to use Samba as a domain controller/active directory? (I am not a windows domain/AD expert nor do I much care to become one). TIA's Pete. -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
[SLUG] Re: Seeking User Group Management Advice
On Tue, Jun 14, 2005 at 10:12:20AM +1000, Alan L Tyree wrote: On Tue, 14 Jun 2005 09:39:16 +1000 Peter Rundle [EMAIL PROTECTED] wrote: SNIP Does anyone know of an open source matlab alternate? The GNU Octave language for numerical computations (2.1 branch) Octave is a (mostly Matlab (R) compatible) high-level language, primarily intended for numerical computations. It provides a convenient command-line interface for solving linear and nonlinear problems numerically. I have not used it and have no idea of its value. It's pretty good for the basics. Last I used it (~18 months ago) it wasn't quite totally Matlab compatible, but there were a surprising number of Matlab libraries that were drop-in compatible, which was nice. Gd sigmonster. Here, have a cookie. - Matt -- A polar bear is a rectangular bear after a coordinate transform. signature.asc Description: Digital signature -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Seeking User Group Management Advice
Alan L Tyree wrote: On Tue, 14 Jun 2005 09:39:16 +1000 Peter Rundle [EMAIL PROTECTED] wrote: SNIP Does anyone know of an open source matlab alternate? The GNU Octave language for numerical computations (2.1 branch) Octave is a (mostly Matlab (R) compatible) high-level language, primarily intended for numerical computations. It provides a convenient command-line interface for solving linear and nonlinear problems numerically. I have not used it and have no idea of its value. I use Octave all the time and think its brilliant. However, Octave is missing all the fancy GUI stuff that Matlab started introducing in Matlab 5. However, the last time I used Matlab was with version 4-something so I don't miss it. Peter, specifically what kinds of things do need? Erik -- +---+ Erik de Castro Lopo [EMAIL PROTECTED] (Yes it's valid) +---+ Religion is a magic device for turning unanswerable questions into unquestionable answers. -Art Gecko, Wombat Discord-1, 128649 -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Seeking User Group Management Advice
Erik de Castro Lopo wrote: I use Octave all the time and think its brilliant. Peter, specifically what kinds of things do need? To be perfectly honest I don't know. I'm not a user of the product I've been asked to deploy it from a central server rather than running a PC version. I just thought I'd ask around and see what's available. Maybe some of the users can use Octave when someone else has locked them out of the licenced matlab. Looks like I need to start checking out Octave. As an advocate of open source I'm trying to encourage it's uptake as much as possible. When you consider that the single user licence of matlab cost more than the hardware (dual xeon) you begin to wonder about the value equation, and it's the public's tax money I'm spending after all. Thanks P. -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Seeking User Group Management Advice
On Tue, 2005-06-14 at 09:39 +1000, Peter Rundle wrote: directory without being able to trash each others account? Is there an elegant way to become another user but retain your group privledge? Perhaps something like login as peter then su - matlab ; newgrp peter? When I do this it prompts for a group password, where is this kept and how do I set it? (man newgrp tells me about the passwd but not how to set it). From what I can see, not many people bother using group passwords, including me. So documentation is fairly sparse. Here be dragons. man group describes how /etc/group is laid out. The second field stores the password. But you'll most likely find that, like user passwords, group passwords are stored in a seperate shadow file, /etc/gshadow , that's only readable by root. The gshadow file on my debian box has passwords disabled for all groups. According to the man page, the root user is able to set passwords on groups using regular old passwd. Check the man page for the appropriate flags. It talks about a group administrator as well, but I'm not sure how you specify that. But root should be able to set a password, and in theory it should work just fine after that. If you decide to chase that any further, I'd like to hear how you go. :-) -- Pete -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Seeking User Group Management Advice
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Tue, Jun 14, 2005 at 09:39:16AM +1000, Peter Rundle wrote: Perhaps something like login as peter then su - matlab ; newgrp peter? When I do this it prompts for a group password, where is this kept and how do I set it? (man newgrp tells me about the passwd but not how to set it). You probably want man gpasswd, the files are /etc/group and /etc/gshadow I remember that there is some reason why no one uses passwords on groups but since I never use them either, I can't remember the reason. Ideally this would all be done in a script so the user clicks on the matlab icon on their icewm desktop (served via ltsp or vnc) and then they key in a passwd or two and volia matlab runs but they retain the ability to open/save work in their home directory. Yes something like that should be possible. You could always teach them to use ftp. Does anyone know of an open source matlab alternate? Depending on your needs, consider R... http://www.r-project.org/ It contains most of the BLASS and LINPACK algorithms and it is good at manipulating large data blocks in memory. Also R has some nice graphics and plotting capabilities. R also has rudimentary symbolic algebra capabilities, it supports expressions and derivatives and substitution into expressions. Since it is designed for statistics, it also has a large statistical analysis library and random number generators. That doesn't mean that it is only good for stats, any time you are manipulating data sets and matricies you need pretty much the same tools. On the other hand, R is not even close to matlab compatible. Personally I think that R has a much nicer high level language interface than what matlab does but if you have a lot of existing code that you need compatibility with then this won't help you. Octave is the best bet if you want close matlab compatibility (but the octave graphics were not much chop last I looked). Both octave and R will run on MS-Windows as well as Linux, I'm pretty sure they both run on Mac OsX too so it should be possible for everyone to work with their code at home before taking it to a big machine for running the more intensive jobs. Mind you, dual xeon isn't big by todays standards anymore, some people probably have bigger machines at home anyhow. This reminds me, the matrix alrogithms can be implemented in single threaded mode or multi-threaded (to take advantage of multiple CPUs). Some large matrix problems will bottleneck at the memory not the CPU so there is no advantage in multiple CPUs (other than a bit more cache), in other cases the multi-CPU is an advantage. Look for a thing called atlas which is here: http://math-atlas.sourceforge.net/ As far as I know neither R nor Octave use atlas natively but you can patch it in and other people have done it (in order to squeeze a bit more performance). For atlas to work properly you have to compile it (and run the self-tuning) on the same hardware as you run it. It tweaks internal buffer chunk sizes to match your system. The Linux server will provide, *nix logins, Imap e-mail, possibly some protected web pages with Apache, and samba shares of their home directory to their desktop PC. I've been considering using LDAP but on past experience found it to be quite a lot to chew, there will only be about 40 users.Any opinions on whether LDAP is worth the effort. I would need users to be able to set their passwords and have that ripple through to each service. Currently PC's are just in a work group, trying to decide whether to use Samba as a domain controller/active directory? (I am not a windows domain/AD expert nor do I much care to become one). If you set the passwords through Samba then it can automatically update the unix login passwords. It is also possible to make both Samba and the unix login authenticate off an existing MS-Windows domain controller if you already have user accounts set up for MS-Windows users (the Samba box has to join the domain, their are instructions on how to do it). - Tel ( http://bespoke.homelinux.net/ ) -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.6 (GNU/Linux) iQIVAwUBQq5LccfOVl0KFTApAQLiqg//TVeek+TO3fxws8jylwF97r6QPz/MXrvN VV2fF7A+U6s9HQFaAs/TBMOnNKpOdE04i8qZUFhdpbfbIpQQVmLhd2L5Hh226O6V quhW5+bGrjdTFJNg8IlEXbJkv6VfnC7yf1R/CEhu+Gw04y5Q60YYT+7ngdxatlTm S+KFVY4Jerpmqsv4Er3ybX0fqbQlFboQHmISogFN+UnheLOxAgfgP99P8DOclUhp DDHq5LUkEXwj8sAJBwpPCSjqVT2Hb2+sVD2Dy/WwFnt3DrYyfet8zXUfvF0ZKuh8 jvUc1uMBdIqs+ywLGQoHQUsyGMWUsV5nM6LNHKL3+siWNPx8pFGNgzmeeEMZRRpw MiUuxfA9oWmdkBaz366KlOGYH0hM8p9liv0uhMiYQhwvQe+gjE6Sz2geza95AwkN 5x6o6ei6PKTaOw23iCtNFlSeBQgF3ZvQQqcXPzJqT/qW/MhPXlO2jik9CuHVYK7c PWLJnzzdSHlffywht9FIp4pwF1Jbd1HqDpUbBpKIgjd7x0YlOaWsghTTuztHR7N9 ntPaazTn+GGlZYO9HXXlHr+YZmuESiHUarsfmBocWDLjiakLP0i7YH/Ykg98ldHK ngNn9G9Pnvd4xrInozz1Pc9/uNn/CEI0Cpr4krZtDgekIewVNJq2uwWowq8iEFCg 8WjE6COqM1s= =xhtY -END PGP SIGNATURE- -- SLUG - Sydney Linux User's Group
Re: [SLUG] Seeking User Group Management Advice
Peter Hardy wrote: From what I can see, not many people bother using group passwords, including me. So documentation is fairly sparse. Here be dragons. Yikes! [snip] If you decide to chase that any further, I'd like to hear how you go. :-) Ok so far I've discovered the gpasswd command which allows me to set a password for the group and apparently apoint an administrator of the group though my idea is to have the user su to the matlab user then newgrp back to their own group (each user being in a group of one). So perhaps I can appoint the user as the administrator of their own group of one at account creation time. So I login as peter, su matlab (without the '-' which means I stay in my directory though my uid/gid is now matlab,matlab) I then $ newgrp peter, which prompts me for the group password and I'm now still in my home directory with uid=matlab, gid=peter. I now run matlab able to read/write files in my own account . When matlab finishes I need to undo the newgrp and su commands I just need it all wrapped up in a script, hmmm. Peter Chub wrote: Use sudo rather than su, and use a wrapper to provide exclusivity. [snip] (Boy three Peters in one thread, who'd thought) Ok, I think I follow, sudo can run a command as another user, cool, That command needs to be newgrp, then matlab, then exit newgrp. Another dragon? I'll follow that line for a bit I thinks P. -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Seeking User Group Management Advice
Hi, --- Peter Rundle [EMAIL PROTECTED] wrote: Does anyone know of an open source matlab alternate? R-project: http://www.r-project.org/ pspp: http://www.gnu.org/software/pspp/pspp.html Regards, SK -- proudly anti-micro$oft -- Shakthi Kannan, MS Software Engineer, Specsoft (Hexaware Technologies) http://www.shakthimaan.com (91) 98407-87007 (M) -- __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
[SLUG] Next ALJ cover disk
Hi all, I'm a contributor to the Australian Linux Journal. We're in the middle of planning the next issue and are trying to decide on a cover disk, and thought it would be good to get the opinions of some Sluggers. So far, we've thought that either a live distro, like Knoppix 3.9, or a full-blown distro, like the latest Debian stable, would make good choices. Our concerns so far are that although Debian 3.1 would be a good choice, it would probably have to be the DVD version, and this might alienate people that still only have CD drives. The flipside of this is that providing just a CD might mean that a lot of packages would need to be downloaded from the Internet. This would be frustrating to dial-up users. Relevancy is also an issue for us. The magazine won't be available for at least another couple of months, so what we pick now will still need to be interesting then. Any thoughts or comments would be gratefully received. :) Mark C. -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Seeking User Group Management Advice
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Tue, Jun 14, 2005 at 01:26:00PM +1000, Peter Rundle wrote: Ok so far I've discovered the gpasswd command which allows me to set a password for the group and apparently apoint an administrator of the group though my idea is to have the user su to the matlab user then newgrp back to their own group (each user being in a group of one). So perhaps I can appoint the user as the administrator of their own group of one at account creation time. So I login as peter, su matlab (without the '-' which means I stay in my directory though my uid/gid is now matlab,matlab) I then $ newgrp peter, which prompts me for the group password and I'm now still in my home directory with uid=matlab, gid=peter. I now run matlab able to read/write files in my own account . When matlab finishes I need to undo the newgrp and su commands I just need it all wrapped up in a script, hmmm. I just tested this on Fedora-3 and noticed that newgrp won't let you change group even when you do have the correct password. I found that RedHat have listed a bug that newgrp is broken here: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=85280 The nasty thing is that newgrp was never updated to support PAM and worse than that, PAM doesn't even accept that group passwords exist. As a result, RedHat have classified this as CLOSED WONTFIX so newgrp is basically broken forever on RedHat. Someone made a comment in the bug report that it works on Suse Linux. It probably works in RedHat if you do NOT use shadow passwords. - Tel -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.6 (GNU/Linux) iQIVAwUBQq5T4sfOVl0KFTApAQLWdRAAjU2o46ltm+9vGmKOOH9ByMFH63TxFlAt gNXMqYgCt78fP0kFmV5nN/cKWlk5ifWVrgMIDwgCa1o8BKA2OmIbrQ6oZNnEqg9+ DRjF5HrGbXr6OKJwQt4bTp2puxuvpACeD2PpJMObNqnlcaT4Ml71joAQxw4b811x DWXdQTHE+ghS1Y3RO1ELM+mfgJBlRHICXHEmKH40IF+bgyogeqrrA/FLt4G9KGev dH0hVhi0fUVQHQsPPQ5hqSyJA/YerSgdovqIlzEFpcboqPQRd7wW6j7HO7I+I4OY lUzcsJuJx1lTZoxgv7auEC7DBHWVJUXeZhPt/u79Xun5YVAhL0YTqwNvO4tcUgCP 4q62Q3dTZ7ODW/x7ZMzTN0VPV21IPSMRUk4AMs8tJhu5sp7i48ArHuZoznxNlqQC Dwhlby0EQisvEp/oGSLIoiP3BTeG8EME5chUviUBSKGy31tlvdJU+CpZQ1QGpBU8 CWtGUCHpmqRrsj1QZnFmS/ki6gcMxtOkYuz5Ud1dCH5f4fHP9MVXp/NWvABM2Wre uB1iTGH0sYSBIe+Zad+05dV9LWX1X9Od8o1tQOeSTr/Ggf1RkbOJk4ST9wMIQOYu vJIi/9dSQYF6LbJ+GXl0hYQr2/Lb/mC+by9m+agcvQb4ujuXUaTxuNYvVU+zH3dY 6rfpsT08Gt4= =nTx4 -END PGP SIGNATURE- -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
Re: [SLUG] Seeking User Group Management Advice
[EMAIL PROTECTED] wrote: I just tested this on Fedora-3 and noticed that newgrp won't let you change group even when you do have the correct password. I found that RedHat have listed a bug that newgrp is broken here: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=85280 The nasty thing is that newgrp was never updated to support PAM and worse than that, PAM doesn't even accept that group passwords exist. As a result, RedHat have classified this as CLOSED WONTFIX so newgrp is basically broken forever on RedHat. Interestingly I just ran this; on FC4 (test2) and it worked like a charm. /home/peter su --command=sg peter /home/matlab/bin/matlab matlab It prompts the user for two passwords which is a bit of a pain but it get's the result that I want. Now I need to chain the matlab account down to a restricted shell, is used to be that you set /bin/rsh as the login shell for the user, but rsh is remote shell now? P. -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html