Linux-Development-Sys Digest #788, Volume #6 Sat, 5 Jun 99 20:14:50 EDT
Contents:
Re: 2.0 <-> 2.2 issues. ("Arkadiy Korobeyko")
Compiling kernel 2.2.x ("Arkadiy Korobeyko")
Re: Terabite Plus Filesystems (Rich Andrews)
Proposal for XML Configuration files (Marco Stipek)
Re: Q: Linux equivalent technologies to Win32? (Peter Eddy)
Re: Proposal for XML Configuration files (Kaz Kylheku)
Re: Function as OLE compound document? (Luigi Montezuma)
Re: Function as OLE compound document? (Roger Larsson)
Re: TAO: the ultimate OS (Stefaan A Eeckels)
Think "Federation" Re: Proposal for XML Configuration files (Christopher B. Browne)
system call (NTVO910)
Re: TAO: the ultimate OS (Christopher B. Browne)
Re: TAO: the ultimate OS ("G. Sumner Hayes")
----------------------------------------------------------------------------
From: "Arkadiy Korobeyko" <[EMAIL PROTECTED]>
Subject: Re: 2.0 <-> 2.2 issues.
Date: Sat, 5 Jun 1999 17:56:07 +0300
Horst von Brand <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> On Tue, 01 Jun 1999 23:11:50 -0400, Omri Schwarz <[EMAIL PROTECTED]> wrote:
> >http://www-stu.calvin.edu/~clug/users/jnieho38/goto22.html
>
> >has a list of updates necessary to go 2.0 -> 2.2,
> >and one of the entries is this gem:
>
> > Kernel modules 2.1.121 (insmod -V to check current version)
> > This is very important! Stuff will not work with earlier versions.
> >Note, however, that once you have 2.1.121
> > installed, don't boot with a 2.0.x kernel or all module hell will
> >break loose. I have no idea why, just trust me.
>
> Wrong. Install 2.1.121 right, it _does_ work with 2.0 kernels (I ran that
> combination for months during the later phases of 2.1). Or use 2.2.2-pre6,
> it also works fine.
I confirm - wrong. Working with kernel 2.0.36-0.7 without any problems.
Arkadiy
------------------------------
From: "Arkadiy Korobeyko" <[EMAIL PROTECTED]>
Subject: Compiling kernel 2.2.x
Date: Sat, 5 Jun 1999 17:53:14 +0300
When trying to enable modules support I receive message:
netsyms: sysctl.wmem.max undeclared here (not in a function)
after which Config exits.
What is wrong?
Arkadiy
------------------------------
From: [EMAIL PROTECTED] (Rich Andrews)
Crossposted-To:
comp.os.ms-windows.nt.admin.misc,comp.sys.sun.admin,comp.sys.hp.misc,comp.os.linux.hardware
Subject: Re: Terabite Plus Filesystems
Date: 5 Jun 1999 17:58:35 GMT
Here is what one can do to create a scalable HA multi terabyte
file system.
1 Sun E4500
1 copy of veritas file system
1 copy of volume mgr
lots of A3500's
configure the controllers with raid mgr to do a simple mirror
then use volume mgr to simple stripe the whole thing together achieving
raid 1+0.
Then use veritas file system so that one does not spend LOTS of time
fscking a file system if it crashes.
One can achieve gigabye per second write speeds this way....granted it was
coupled with a e10k..but that is another story.
rich
------------------------------
From: Marco Stipek <[EMAIL PROTECTED]>
Crossposted-To: comp.unix.programmer,comp.os.linux.development.apps
Subject: Proposal for XML Configuration files
Date: Fri, 04 Jun 1999 18:33:43 +0200
Reply-To: de.comp
Linux/Unix is one of the best operating systems worldwide, but in my
opinion
there is one bad thing thousends of config files in any format you can
imagine,
changing from version to version (like old and new Bind configs).
Is there anybody in the web thinking about a standard on this topic,
maybe based on XML?
If not maybe there is any developer interested in starting such a
workgroup. Get touch with me.
Marco
------------------------------
From: Peter Eddy <[EMAIL PROTECTED]>
Subject: Re: Q: Linux equivalent technologies to Win32?
Date: Sat, 05 Jun 1999 13:41:26 -0400
George Craig wrote:
>
> For example, what is the standard Linux SDK/technology for building
> distributed applications? Is it CORBA?
CORBA is one standard, probably the best bet if you're looking for
something similar to COM. Other options include RPC (not very COM like
at all) and DCE (cringe) and of course ObjectSpace Voyager, which is
Java.
> Listed below are additional technologies that I have worked with and would
> like to have compared to the equivalent *unix* incantation.
>
> 1. COM / ActiveX (is it CORBA?)
> 2. DCOM (is it RPC, RMI in Java?)
All CORBA implementations are able to work in a distributed manner, at
least I've never seen one that couldn't. Here's good link to get
started: http://patriot.net/~tvalesky/freecorba.html
RPC, Java RMI, Java+CORBA, all possibilities.
> 3. Transaction Monitoring of objects (Enterprise JavaBeans - ETS?)
There are various implementations for Unix, including BEA Tuxedo. Of
course EJBs are very nice and I imagine will become very widely used.
> 4. Win32 API (Any Window Manager flavor API and Libs? i.e., Gnome)
You have more options here than you probably want. Here's a very nice
place that compares various options: http://www.free-soft.org/guitool/
Hope that helps, and welcome to the Unix world!
Peter
------------------------------
From: [EMAIL PROTECTED] (Kaz Kylheku)
Crossposted-To: comp.unix.programmer,comp.os.linux.development.apps
Subject: Re: Proposal for XML Configuration files
Date: Sat, 05 Jun 1999 18:53:35 GMT
On Fri, 04 Jun 1999 18:33:43 +0200, Marco Stipek <[EMAIL PROTECTED]> wrote:
>Linux/Unix is one of the best operating systems worldwide, but in my
>opinion
>there is one bad thing thousends of config files in any format you can
>imagine,
>changing from version to version (like old and new Bind configs).
Of course, the only worse thing is thousands of registry keys whose
values change from version to version.
Get real, config files rule. The more syntaxes the merrier.
>Is there anybody in the web thinking about a standard on this topic,
>maybe based on XML?
The nice thing about standards is taht there are so many to choose from.
For example, you could use the X resource stuff, in which case
your config files would end up looking like:
! comment
foo.bar.xyzzy: value
>If not maybe there is any developer interested in starting such a
>workgroup. Get touch with me.
Good luck convincing everyone to use whatever scheme you come up with.
------------------------------
From: Luigi Montezuma <[EMAIL PROTECTED]>
Crossposted-To:
comp.object.corba,comp.os.linux.development,comp.os.linux.development.apps
Subject: Re: Function as OLE compound document?
Date: Sat, 05 Jun 1999 15:50:29 -0700
Enosh Chang wrote:
> ...is there any function like OLE to generate
> compound document in Linux?
No.
==========================================
len
if you must use email, reply to:
73 662 dot 26 51 at compuserve dot com
------------------------------
From: Roger Larsson <[EMAIL PROTECTED]>
Crossposted-To:
comp.object.corba,comp.os.linux.development,comp.os.linux.development.apps
Subject: Re: Function as OLE compound document?
Date: Sat, 05 Jun 1999 21:09:19 GMT
Yes,
Check out KOM/OpenParts.
http://developer.kde.org/openparts/html/openparts.html
KOffice is being written with it. More about KOffice at
http://koffice.kde.org
Regards,
Roger Larsson
Luigi Montezuma wrote:
>
> Enosh Chang wrote:
> > ...is there any function like OLE to generate
> > compound document in Linux?
> No.
>
> ------------------------------------------
> len
> if you must use email, reply to:
> 73 662 dot 26 51 at compuserve dot com
--
The Internet interprets Windows as damage,
and routes around it.
Roger Larsson
Skellefte�
Sweden
------------------------------
From: [EMAIL PROTECTED] (Stefaan A Eeckels)
Crossposted-To: alt.os.linux,comp.os.linux.advocacy,comp.os.misc,comp.unix.advocacy
Subject: Re: TAO: the ultimate OS
Date: 5 Jun 1999 21:22:09 GMT
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Vladimir Z. Nuri) writes:
> Jimen Ching ([EMAIL PROTECTED]) wrote:
>
>: 1. The first mistake you made was to umbrella the OS concept. An operating
>: system is more than just what the user sees. The idea that an OS should be
>: designed from the ground up to cater to the end-user is looking at only half
>: the picture.
>
> its the whole picture, man. that's the point. the end user is the
> point. linux developers currently refuse to recognize this. they
> think the OS designers are part of the equation. in the end
> the OS creators (hackers/programmers) must be irrelevant in the face of the
> end user.
IMHO it's the fundamental flaw in your concept. If you really want
to think about end-users, you should think (as has been pointed out)
about a new HCI. The end-user (as the success of Windows so clearly
proves) is totally unaware of the technology underlying the system
he uses.
Once you have a HCI, *then* you can evaluate whether you need a new
toolkit (a la qt/KDE), a new windowing system (a la X, if your HCI
still uses the WIMP approach), an alternative to DCOM/CORBA, a different
file system paradigm, or (and I doubt it) a new OS or a new
hardware architecture.
--
Stefaan
--
PGP key available from PGP key servers (http://www.pgp.net/pgpnet/)
___________________________________________________________________
Perfection is reached, not when there is no longer anything to add,
but when there is no longer anything to take away. -- Saint-Exup�ry
------------------------------
From: [EMAIL PROTECTED] (Christopher B. Browne)
Crossposted-To: comp.unix.programmer,comp.os.linux.development.apps
Subject: Think "Federation" Re: Proposal for XML Configuration files
Reply-To: [EMAIL PROTECTED]
Date: Sat, 05 Jun 1999 21:57:27 GMT
On Fri, 04 Jun 1999 18:33:43 +0200, Marco Stipek <[EMAIL PROTECTED]> posted:
>Linux/Unix is one of the best operating systems worldwide, but in my
>opinion
>there is one bad thing thousends of config files in any format you can
>imagine,
>changing from version to version (like old and new Bind configs).
>
>Is there anybody in the web thinking about a standard on this topic,
>maybe based on XML?
>
>If not maybe there is any developer interested in starting such a
>workgroup. Get touch with me.
Unfortunately, format is only one of several issues, and is arguably
not even the important one.
Be aware that nothing mandates that XML information be represented
as XML text; it may very well be more appropriate to store the data
in some other form, perhaps in a database, with XML "merely" being
the public face the data puts on when you try to use it as a form for
data interchange.
The use of a particular format as the physical expression does not
mean that you have a solution to such issues as ensuring that you can
safely update the data. If there might be multiple processes with access
to the data, that might be accessing it across NFS, the issue of having
a suitable format does nothing to resolve access issues.
The primary problem with using XML as a "database-like" form is that it
is inherently (and arguably intentionally) rather fragile. That is, XML
documents are intended to be validatable, and the results of processing
an invalid document are not defined. If someone inserts a bad line of
data into a document, it is perfectly acceptable behaviour (from at least
a theoretical perspective) to reject the document forthwith.
Obviously that's not too friendly an action to take, but it is necessary
to "beg the question" of what *should* be done if the quality of the
database degrades. I would argue that the format is fragile, and while
it is great for interchange of already-serialized data, it is not so
great for storage/updates.
In a sense, XML is as "tempting" an option as the absolute opposite,
the *also fragile* "Windows Registry," where everything gets dropped
into a large, fragile binary database structure.
They represent opposite ends of the spectrum in that a "big XML database"
represents a file that can't be modified without rewriting the whole
thing, whereas the "Registry" allows easy modification of any key anywhere
at any time. (Really rather like having everything stored in a DBM file.)
It's quite clear that XML versus DBM have *very* different characteristics
when it comes to things like locking. Both are extreme approaches that
have big strengths and *huge* weaknesses that get steadily more dramatic
the more extensive the population of configuration data that you push
into them.
Importantly, from the perspective of a would-be programmer-user of
such a configuration system, the value of a configuration management
system is *not* in what format it uses, but rather in whether the
system makes it convenient to store both singleton values as well as
"somewhat structured data" without making the programmer implement a
whole lot of locking logic to ensure that updates are handled safely.
The configuration system should manage that *for* the programmer, which
would make it a valuable "service."
The other major thing for a configuration management system to provide
is the ability for programmers to not need to worry a whole lot about
parsing out data files.
A libDBM or a libXML can provide that benefit; so can a libINI or a
libPropList (hopefully it should be obvious that these would respectively
work with DBM, XML, "Windows-style .ini" and OpenStep-style directory
structures).
The thing that it's not clear that these all provide is a way for
some common metadata to be stored that would be useful to the central
"configuration manager" program so that diverse sets of config information
could be joined together as a federation of configuration databases.
I'm taking the term "federation" both from Michael Stonebraker's recent
database work (the Mariposa system, commercialized as Cohera) and from
the CORBA notion of a combination of services accessible by a group of
Trading Service servers.
The point here is that the *really* smart idea is to take some of
the existing configuration schemes (that are mature enough that we
know they're pretty robust), and tweak them just enough as to allow
them to communicate with the central process that would build a sort
of confederation out of them. This doesn't forcibly require One True
Data Format. It's obviously a nice idea if new applications that get
constructed use relatively similar configuration schemes that are easy
to get integrated into "The Federation."
--
Those who do not understand Unix are condemned to reinvent it, poorly.
-- Henry Spencer <http://www.hex.net/~cbbrowne/lsf.html>
[EMAIL PROTECTED] - "What have you contributed to free software today?..."
------------------------------
From: [EMAIL PROTECTED] (NTVO910)
Subject: system call
Date: 5 Jun 1999 21:46:13 GMT
Ok, I want my program to use the "su" or some similar command to change the
account that the user is on. I want it to try to change to the root account
and type in the password for the user.
If the user is doing it, it's done like this:
________________
[EMAIL PROTECTED]> su
Password: --password goes here--
________________
duh, ok. So I checked the manuals to try and find some command like change
account or something but the command line "su" was all that I found. I tried
to use:
system("su");
system(password_ptr);
However, after the program uses "su," it just waits for the response and does
not right the data in password_ptr to the prompt. Do I have to use some other
function? Do I have to do it some little different way? Am I a complete
idiot?
Ok so I am.
Thanks for all help.
Braden
------------------------------
From: [EMAIL PROTECTED] (Christopher B. Browne)
Crossposted-To: alt.os.linux,comp.os.linux.advocacy,comp.os.misc,comp.unix.advocacy
Subject: Re: TAO: the ultimate OS
Reply-To: [EMAIL PROTECTED]
Date: Sat, 05 Jun 1999 22:32:19 GMT
On 5 Jun 1999 21:22:09 GMT, Stefaan A Eeckels <[EMAIL PROTECTED]> posted:
>In article <[EMAIL PROTECTED]>,
> [EMAIL PROTECTED] (Vladimir Z. Nuri) writes:
>> Jimen Ching ([EMAIL PROTECTED]) wrote:
>>
>>: 1. The first mistake you made was to umbrella the OS concept. An operating
>>: system is more than just what the user sees. The idea that an OS should be
>>: designed from the ground up to cater to the end-user is looking at only half
>>: the picture.
>>
>> its the whole picture, man. that's the point. the end user is the
>> point. linux developers currently refuse to recognize this. they
>> think the OS designers are part of the equation. in the end
>> the OS creators (hackers/programmers) must be irrelevant in the face of the
>> end user.
>IMHO it's the fundamental flaw in your concept. If you really want
>to think about end-users, you should think (as has been pointed out)
>about a new HCI. The end-user (as the success of Windows so clearly
>proves) is totally unaware of the technology underlying the system
>he uses.
>Once you have a HCI, *then* you can evaluate whether you need a new
>toolkit (a la qt/KDE), a new windowing system (a la X, if your HCI
>still uses the WIMP approach), an alternative to DCOM/CORBA, a different
>file system paradigm, or (and I doubt it) a new OS or a new
>hardware architecture.
If the goal is to have a new OS, then it's just plain stupid to preach
to Linux folk. It would be *far* more fruitful to examine alternatives.
If Vladimir were as cognizant of OO as he seems to think, he *should*
figure it wiser to look at FluxOS, an OS designed to allow diverse things
to be connected in, or perhaps Hurd, where recent discussions appear to
be leading towards having an ORB that sits at the Mach level.
Those are two better ways of getting into a "truly OO system" than
preaching at groups of people who, as likely as not, may actually *like*
the way UNIX is.
Of course, that would require becoming more acquainted with the
literature of operating system research. There's quite a lot of stuff
not out on the web, but search engines *are* your friend...
--
Those who do not understand Unix are condemned to reinvent it, poorly.
-- Henry Spencer <http://www.hex.net/~cbbrowne/oses.html>
[EMAIL PROTECTED] - "What have you contributed to free software today?..."
------------------------------
From: "G. Sumner Hayes" <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.advocacy,comp.os.misc,comp.unix.advocacy
Subject: Re: TAO: the ultimate OS
Date: Sat, 05 Jun 1999 19:10:05 -0400
"Vladimir Z. Nuri" wrote:
> the essay is basically suggesting a c++ like object
> is what is preferred.
That's what I suspected, which is why I kept saying things like
"You never say what an object is. If you mean bundling of code
and data, here's why it's considered BAD as the atomic abstraction
for developing UIs". That object model assumes a 1-1 association
between code and data. That's great for certain things, but really
bad for UI design. You can use that object model as a low-level
building block inside a well-designed UI, for sure; using it as the
sole high-level abstraction is a really poor idea, as any
introductory HCI text will surely mention. Most useful UIs need to
be concerned about data presentation and manipulation, which almost
mandates a decoupling of data and code when you're talking about
the high-level view.
That's the major point I've been trying to make. Any system that
thinks that strongly-typed code/data bundled objects are a good
atomic high-level abstraction for building UIs is going to come up
with something that's markedly worse than most modern UI designs.
Nobody who designs UIs for a living would try to use that kind of
object as the atomic abstraction. Your paper presents it as the
sole abstraction -- "everything's an object" -- which is going
to lead to major heartache. It's a huge step _backward_ from
"everything's a file", which isn't exactly a perfect abstraction
itself.
Anyway, I'm tired of discussing this topic. All of this would
be covered in intro HCI or UI design textbooks, which is what
I've been advocating you read from the start.
> do you go and harangue everyone in object newsgroups about the defn
> of "object" every time they use the term
No. In comp.object.* the posters almost always use a Subject: line
or have something in the body that indicates what sort of object
system they're dealing with.
If you post something like "Is object-oriented programming better
than functional programming?" the first thing most of the readers
there will say is "What do you mean by object-oriented?" and the
second is "better at what?".
You should probably read the comp.object FAQ. The first question
in it is "What is an object?". It gives an overview of some common
definitions. It also has interesting tidbits like:
: 1.16) Is A Class An Object?
: ----------------------------
:
: In C++ no[SNIP]
The explanation makes no sense out of context, so you should read
through the whole FAQ if you're interested. Note that the quote
does not say C++ isn't an OO language, it just says that a class in
C++ is not an object.
--Sumner
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.development.system) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Development-System Digest
******************************
