Linux-Development-Sys Digest #903, Volume #7 Wed, 24 May 00 20:13:15 EDT
Contents:
Re: Need ideas for university funded project for linux (Praedor Tempus)
Re: Need ideas for university funded project for linux (Matthias Warkus)
Re: please help on installing FA310 network card(from Netgear) (Warren Young)
Re: Need ideas for university funded project for linux ([EMAIL PROTECTED])
Re: Need ideas for university funded project for linux ([EMAIL PROTECTED])
Re: Need ideas for university funded project for linux ([EMAIL PROTECTED])
unresolved symbol pmd_offset - huh? (Timur Tabi)
Re: Need ideas for university funded project for linux (Leslie Mikesell)
Re: how-to develop? (Nigel Kukard)
Re: 970Cxi - HP refuses to help customer (Dima Maziuk)
Re: unresolved symbol pmd_offset - huh? (Daniel R. Grayson)
----------------------------------------------------------------------------
From: Praedor Tempus <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: Wed, 24 May 2000 15:24:03 -0600
David Steuber wrote:
>
> [EMAIL PROTECTED] (Dowe Keller) writes:
>
> ' [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> ' >David Steuber <[EMAIL PROTECTED]> writes:
> ' >
> ' >> Do people really have trouble with ./configure, make, make install?
> ' >> It has _never_ been a problem for me. Maybe I am just lucky. Even
> ' >> though I changed my compiler, libc, and libtools.
> ' >
> ' >That precise process usually works out fine. However, a number of
> ' >these processes require manual modification of the Makefile or a
> ' >custom configuration file. I've also encountered several configure
> ' >scripts that break, and when that happens, you're doomed to rewriting
> ' >the Makefile by hand. And there are still a few programs that just
> ' >provide you with a grab-bag of Makefiles, and you get to pick which
> ' >one you want. Those are *always* disasters, but usually the Makefiles
> ' >are at least short enough that fixing them isn't impossible.
> '
> ' Yow, you make it sound like brain surgery. I can count the number of
> ' times that I had to hack Makefiles to get a program to make on the
Ah, then you are fortunate, or you are using a subclass of the software
I use. I constantly run into problems with the "simple" ./configure,
make, make install process. It usually derives from ./configure not
doing its job. Case in point: I just tried compiling i3d, a 3d
graphics modeller using ./configure, etc. It doesn't work. Nor has
3dom, nor Mindseye and a host of other apps I've compiled. I would
say that the failure rate is in the range of 40% which sucks big.
With i3d, configure runs and makes all the makefiles at the end.
"Ah good", I think, and then run make. It chugs away for a bit
and finally craps out over qtgl (which IS installed) because it
says it isn't there. Ahem...if something required for the making
of the source wasn't there, then configure should have found this
out and not gotten far enough to write the makefiles. At least
with RPM installs I get a pointed list of failed dependencies that
I can easily fix. With the ./configure game it isn't that simple.
I often DO have the header or lib installed that it is failing to
find and I end up having to tweak the Makefile or experiment with
./configure switches to get it to work (about 60% of the time).
The remaining 40% of the 40% that fail I generally give up on
and look for a precompiled binary since there is something subtle
screwy with what the app wants and what I have...and ./configure
doesn't cut the custard. It is failing to find the problem.
I do not automatically check the config.log EVERY time I run
./configure. I should only have to if there is a configuration
failure (and such failure should prevent the Makefiles from being
generated).
Either config scripts need to be better written or a better system
needs to be developed.
I haven't used deb packages so I cannot address them but it would be
nice if when installing an rpm (or trying to build a source rpm) and
a dependency fails, then you should be given the automatic option of
downloading the missing rpm rather than just getting a name. How
about something that follows this script:
========
You: "rpm -Uvh <someapp.rpm>"
This rpm requires <somedependency>
Would you like to download and install the required package(s)? [y/n]
Connecting to...(your preferred ftp or http site or selected from a good
pre-created but editable list or an option to pull from a CD or
harddrive)
Downloading > ######### (hashing ala installing rpm with -Uvh)
Installing dependency
################ (hashing)
Installing <someapp.rpm>
=======
There. Done. Little pain or muss and the deed is done without a lot of
dicking
around. You are up and using the app rather than trying to work out why
you
can't use it or manually searching for the missing rpms. It would be
nice as
well if when picking up the dependency rpm it would check to see if IT
has any
particular dependencies to satisfy...
A graphics frontend that can do this would also be nice. Perhaps
kpackage or
the gnome equivalent could be improved to do this.
Something similar could be done for src.rpms that die for missing
dependencies (I run into this periodically but far less than the problem
arises with ./configure && make && make install).
praedor
------------------------------
From: [EMAIL PROTECTED] (Matthias Warkus)
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: Wed, 24 May 2000 17:17:50 +0200
Reply-To: [EMAIL PROTECTED]
It was the Wed, 24 May 2000 13:00:00 GMT...
...and David Steuber <[EMAIL PROTECTED]> wrote:
> [EMAIL PROTECTED] (Matthias Warkus) writes:
>
> ' It was the Tue, 23 May 2000 08:59:59 GMT...
> ' ...and David Steuber <[EMAIL PROTECTED]> wrote:
> ' > The right is non-exlusive. That means everyone can get that right. I
> ' > think TrollTech is just trying to prevent forking of the Qt library
> ' > here.
> '
> ' Exactly that is which is bad IMHO. Real software freedom has always
> ' been the freedom to fork.
>
> That's a good point. But what is the value in real forking?
- Sometimes, forking is good. Look at XEmacs vs. GNU Emacs. It's good
that we've got both of them, isn't it?
- Better code reuse. Just an example: you've got an FPPLed[0] image
manipulation program with a really neat and fast algorithm to
rasterise cutesy little heart shapes. You want to have this
functionality in an Imlib-like library. You can't because you'd have
to fork the code off to modify it and redistribute the "modified
version" as a library.
Now, you might argue this is not forking, and by some clever
legalese, one can make code reuse legal and forking illegal. But
where do you draw the line?
- Enabling people to do the Real Thing(TM). Assume that some company
develops the FPPLed software package FooPro 2000(TM) which is really
cool and useful. Unfortunately, its successor, FooPro 2001(TM), is a
crock, it's taken a design road that you absolutely hate. There are
some improvements you want to see in FooPro some day, but if the
official roadmap of FooPro goes on that way, the future versions
will be horrible.
What are you going to do? Persuade the company to incorporate your
patches and make their design better? Just go with the old FooPro
2000? You can't just patch FooPro 2000 and release your own version
-- forking prohibited. Of course you could release megabytes of
patches and put them on your FTP server next to the original FooPro
2000 source, but well...
You see, if you can't fork it, the software will effectively remain
controllable by the copyright holder(s).
> Do you really want to have ten different major versions of GTK+
> floating around? Or even two? If an application says it uses GTK+
> ver x.y, wouldn't it be simpler to know that the application didn't
> really mean FGTK+ x.y?
Try to find evidence for just *one* fork of that kind. Most forks are
rather benign (like XEmacs/GNU Emacs). I don't know whether any
libraries with established and popular APIs ever forked.
> Linux is held together because people respect the opinion of Linus
> Torvalds. Even so, there are Alan Cox diffs, RTLinux, and probably
> others. The tendency is to stick with Linus Torvalds Linux as the
> base. Will that be true for all GPL projects? What if some group
> decides a certain feature is needed in GTK+, but another group of
> equal size feels that feature doesn't belong, or should be implemented
> in a different way, with a different interface?
This is exactly the point: A free software project always depends on a
consensus. If the consensus is lost, the project splits. The software
has got to split along with it, or those people splitting away from
the main branch would lose their rights w/r/t the software package if
they weren't allowed to fork off their version.
mawa
[0] Fork-Protected Public Licence
--
Focus-Leser!
Dankesager!
Defensivfahrer!
Handschuhtanker!
------------------------------
Date: Wed, 24 May 2000 15:32:40 -0600
From: Warren Young <[EMAIL PROTECTED]>
Subject: Re: please help on installing FA310 network card(from Netgear)
chuliang Yu wrote:
>
> gcc -DMODULE -D__KERNEL__ -I/usr/src/linux/net/inet -Wall
> -Wstrict-prototypes -O6 -c tulip.c
This is not required with current kernels. They have "tulip" support
built into them. See your distro manual on how it wants to be told
about a new network card.
--
= Warren -- ICBM Address: 36.8274040 N, 108.0204086 W, alt. 1714m
------------------------------
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: Need ideas for university funded project for linux
From: [EMAIL PROTECTED]
Date: Wed, 24 May 2000 21:42:25 GMT
[EMAIL PROTECTED] (Leslie Mikesell) writes:
> In article <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
> >David Steuber <[EMAIL PROTECTED]> writes:
> >> If Windows is so great, why do you have to reboot when you change your
> >> IP address?
> >You don't.
> For 1 value of windows.
Granted. But they've at least recognized that it's a problem and have
fixed it.
> Why do you have to reboot when you change the machine name?
In an NT/2K environment, probably because you're logged in via the
server service. So if you login to APATHOS as Administrator, you're
APATHOS\Administrator (even if it's just a workgroup instead of a
domain). So when you change the machine name, your login becomes
invalid and authentication will fail. Even if you're not logged in
*over* the network, but through a loopback connection or even just
locally, the system still makes it look like you're signed on over the
network.
I don't know this for a fact, but it certainly makes sense to me.
Now, you may want to make a case for the brokenness of SMB, but if I'm
right, then Windows is at least doing it properly.
--
Eric P. McCoy ([EMAIL PROTECTED])
non-combatant, n. A dead Quaker.
- Ambrose Bierce, _The Devil's Dictionary_
------------------------------
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
From: [EMAIL PROTECTED]
Date: Wed, 24 May 2000 22:02:05 GMT
"Anthony W. Youngman" <[EMAIL PROTECTED]> writes:
> >> I am led to believe (in other words I may well be wrong...) that rpms
> >> basically have a required/not-required status. If the system MAY require
> >> a package, then either it is flagged as required and the system tries to
> >> make you install it, or it's not flagged and gets ignored.
> >Well, technically, some is either required or it isn't. If you're
> >right (I have no idea), the problem seems to be more on the package
> >maintainer's end, rather than the rpm developer's end.
> But you're ignoring the example packages I (deliberately) chose ...
> If I have an ISDN card, then I *NEED* ISDN4LINUX, if I don't then it's a
> waste of space. Same with a sound card and OSS.
You *need* it? As in, you won't be able to use your system at all
unless you have it?
I'm deliberately choosing a black-and-white perspective here, because
you're saying that's what rpm has. But it's not totally unjustified;
raw ISDN support is done in the kernel, correct? If so, should the
package manager communicate with the kernel to determine if a package
should be required?
Really, it seems clear to me that ISDN4LINUX and OSS are not
required. They are useful if you have the appropriate hardware, but
if not having them doesn't break your system, then they just aren't
needed. I can think of lots of reasons not to use OSS even if you
have a sound card. The case is weaker for ISDN - because no
computers, to my knowledge, ship with it installed - but maybe one
could still be made.
Note that your suggestion (marking something "optional") is
practically identical to marking something not-required, except for
package grouping. Optional packages aren't installed by default, so
you have to select them manually. Not-required packages aren't
installed by default, so you have to select them manually. The same
degree of effort is involved in both.
> If that hardware is present, then those packages are REQUIRED. If the
> hardware isn't there then those packages are a waste of space (and on
> the system I was complaining about, it was more than 1% of the available
> disk space for ISDN4LINUX alone - that's space I can't spare).
Well, you can always remove them. I see your complaint, though.
> As somebody else pointed out, rpms can't have conditional dependency.
> Either it's flagged as "required" and I scream blue murder because the
> basic install on my mum's pc crashes with a "disk full", or it's flagged
> as "not required" and I scream blue murder because it doesn't install on
> the office server and I need it.
I'm going to have to disagree with you here, just on technical
reasons. To implement the kind of conditional dependency you want
(which Debian tries to do and screws up royally [1]), you'd need help from
the kernel. It could certainly be done, but I don't like the idea of
tying rpm or dpkg or whatever to a specific kernel.
Perhaps a separate program which *is* tied to a specific kernel and
reports on what hardware is installed? It'd be really simple; its
only job would be to get /proc/pci and friends in an
architecture-independent format. This could also be used in the
install process to autoselect some modules the user is likely to want.
--
Eric P. McCoy ([EMAIL PROTECTED])
non-combatant, n. A dead Quaker.
- Ambrose Bierce, _The Devil's Dictionary_
[1] Having two separate "enlightenment" and "enlightenment-nosound"
packages that differ only in their dependencies is, in my opinion,
broken packaging.
------------------------------
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
From: [EMAIL PROTECTED]
Date: Wed, 24 May 2000 22:05:03 GMT
[EMAIL PROTECTED] (JEDIDIAH) writes:
> Actually, Linux was forked several times in the process
> of moving to new platforms and continues to be. Linus
> himself even considers this a 'good thing'.
Because it usually gets merged back to the "official" Linux by the
next [0-9]\.[02468] release.
--
Eric P. McCoy ([EMAIL PROTECTED])
non-combatant, n. A dead Quaker.
- Ambrose Bierce, _The Devil's Dictionary_
------------------------------
From: Timur Tabi <[EMAIL PROTECTED]>
Subject: unresolved symbol pmd_offset - huh?
Date: Wed, 24 May 2000 22:12:30 GMT
I'm trying to get a test module to work, and I'm seeing an error I can't
explain. I'm using 2.3.99-pre6 on a Pentium III. Here's the relevant
code:
int global = 0x12345678;
int init_module(void)
{
unsigned long phys = virt_to_phys(&global);
pgd_t *pgd = pgd_offset(mm, (unsigned long) &global);
pmd_t *pmd = pmd_offset(pgd, (unsigned long) &global);
pte_t *pte = pte_offset(pmd, (unsigned long) &global);
...
And when I do "insmod test.o", I get this:
test.o: unresolved symbol pmd_offset
test.o: unresolved symbol virt_to_phys
Now, I know that the PMD is "optimized out" on Intel platforms (I'm not
really sure what that means), but I can't understand why this doesn't
work when I see lots of very similar code in other modules that does
work.
I'd also like to know why virt_to_phys can't be found. BTW, this code
compiles without any errors.
--
Timur "too sexy for my code" Tabi
Remove "nospam_" from my email address when replying
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: [EMAIL PROTECTED] (Leslie Mikesell)
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: 24 May 2000 17:20:24 -0500
In article <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
>> If that hardware is present, then those packages are REQUIRED. If the
>> hardware isn't there then those packages are a waste of space (and on
>> the system I was complaining about, it was more than 1% of the available
>> disk space for ISDN4LINUX alone - that's space I can't spare).
>
>Well, you can always remove them. I see your complaint, though.
This kind of stuff really should be done at boot-time, since you
can always install the card after you install the system. Won't
kudzu notice and do something sensible?
Les Mikesell
[EMAIL PROTECTED]
------------------------------
Date: Thu, 25 May 2000 00:38:45 +0200
From: Nigel Kukard <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: how-to develop?
Nicola Attico wrote:
> Hello,
>
> I'm serching for a word of advice.
> I'm searching for a project to work at
> (starting from zero [maybe better] or
> running).
> The point is that I've not strong C
> knowledge, so I would like to learn...
> Anyway I'm not a total unable
> and I've years of experience using
> Linux and some practice in administration...
> I think moreover that the better way
> to learn how to write software is to write
> it, so I would like to learn doing it!
> So, I'm searching for a *baby* project,
> where it is possible to learn...
>
> Does it exists?
>
> Thanks,
>
> Nicola
Are you looking for a job or just summin to do to help u learn? :-)
Regards
Nigel
------------------------------
From: Dima Maziuk <[EMAIL PROTECTED]>
Subject: Re: 970Cxi - HP refuses to help customer
Date: Wed, 24 May 2000 18:48:19 -0500
Nicolas Eymerich wrote:
>
> Anybody knows if is possible to install a parallel port spy utility to
> extract the control code used?
Dunno but you can capture parallel port thruput with dosemu. I tried
that on a dongle once...
--
(1) Office employees will daily sweep the floors, dust the
furniture, shelves, and showcases.
(2) Each day fill lamps, clean chimneys, and trim wicks.
Wash the windows once a week.
(3) Each clerk will bring a bucket of water and a scuttle of
coal for the day's business.
(4) Make your pens carefully. You may whittle nibs to your
individual taste.
(5) This office will open at 7 a.m. and close at 8 p.m. except
on the Sabbath, on which day we will remain closed. Each
employee is expected to spend the Sabbath by attending
church and contributing liberally to the cause of the Lord.
-- "Office Worker's Guide", New England Carriage Works, 1872
------------------------------
Subject: Re: unresolved symbol pmd_offset - huh?
From: [EMAIL PROTECTED] (Daniel R. Grayson)
Date: 24 May 2000 19:06:56 -0500
Perhaps you are including this definition from pgtable-2level.h (check this.):
extern inline pmd_t * pmd_offset(pgd_t * dir, unsigned long address)
{
return (pmd_t *) dir;
}
If you give options to gcc that allow it to "inline" it, then your file will
contain no unresolved reference to it. Such options include
-finline-functions
-O3
------------------------------
** 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
******************************
