Linux-Development-Sys Digest #342, Volume #8 Tue, 12 Dec 00 15:13:13 EST
Contents:
MS Project equivalent ("Ping-Ya Ko (������)")
Making a new file system (Sean)
Re: Compiling C++ programs with GCC --> no GPL license implications (Ken Arromdee)
Re: How to optimize the Kernel-Compile ([EMAIL PROTECTED])
Re: Compiling C++ programs with GCC --> no GPL license implications (Pete Becker)
source code of mkfs.minix ("Marty")
Device Driver ("Mike Austin")
Re: changing BASH's path searching (Alex Graf)
Re: Compiling C++ programs with GCC --> no GPL license implications
([EMAIL PROTECTED])
Re: Compiling C++ programs with GCC --> no GPL license implications
([EMAIL PROTECTED])
Re: source code of mkfs.minix ([EMAIL PROTECTED])
Re: slow asymmetric half-duplex communication (ADSL) (Dave Platt)
Re: Compiling C++ programs with GCC --> no GPL license implications (Pete Becker)
Re: imaginary complex value ([EMAIL PROTECTED])
kernel questions ([EMAIL PROTECTED])
----------------------------------------------------------------------------
From: "Ping-Ya Ko (������)" <[EMAIL PROTECTED]>
Subject: MS Project equivalent
Date: Tue, 12 Dec 2000 07:21:25 +0800
Hi everyone,
Is there a MS Project equivalent application in Linux? It will be very
exciting if there is one. Therefore, we can manage project through internet.
Thanx a lot
pyko
------------------------------
From: Sean <[EMAIL PROTECTED]>
Subject: Making a new file system
Date: Tue, 12 Dec 2000 23:23:07 +0800
I am trying to build a new file system based on Minix !
Are there any registration in the kernel needed to be done to support
one more file system ?
------------------------------
From: [EMAIL PROTECTED] (Ken Arromdee)
Crossposted-To: comp.lang.c++,gnu.misc.discuss
Subject: Re: Compiling C++ programs with GCC --> no GPL license implications
Date: 12 Dec 2000 15:26:23 GMT
In article <[EMAIL PROTECTED]>,
Pete Becker <[EMAIL PROTECTED]> wrote:
>> Tisdale is a known troll. Save your keystrokes.
>Bad idea. He happens to be right. Rights under copyright law are decided
>by statues and by courts, not by public announcements.
*Giving up* a right can be done by public announcements.
--
Ken Arromdee / [EMAIL PROTECTED] / http://www.rahul.net/arromdee
"Eventually all companies are replaced." --Bill Gates, October 1999
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: How to optimize the Kernel-Compile
Date: Tue, 12 Dec 2000 14:57:37 GMT
[EMAIL PROTECTED] wrote:
> There were many groups with different opinions what optimal code for
> i386 and descendants is, there is the gcc crowd, the pgcc crowd, the
> kernel hackers, and other people. They didn't come to the same
> conclusions, so there are different ways to get optimal results in
> respect of one or another hacker crowd. The kernel is especially
The problem, of course, being that optimal is quite a moving
target. The kernel aside for a moment, there are a depressing number
of people who insist on applying every optomising flag in the gcc
manual to generate the "_best_ code possible" with no regard for the
reliabilty problems that may come from churning things into spaghetti
or the fact that a .002% increase in speed for a program whith a
runtime under a minute is a really meaningless increase. I mean,
pentium-optimised distributions may be a nice idea, but do you really
need an optomised ls?
Back on the kernel, lord knows I have issues with portions of it, by
this point in time who doesn't? I remember installing from a pair of
5.25" floppies, and compiling it in minutes on a slow 486. Since then,
of course, many features have gone in which I don't particuarly
like. I suspect alot of people have the same basic experience.
One thing I don't question though, is that Linus in particular, and
the other developers in general are commited to optomising the
kernel. They've made some wrong turns here and there, and it's a
lively debate at times, but as a general rule, performance increases
always make it in. If there are additional portions that could benefit
from different flags on the 486 (and by benefit, I mean actually show
some sort of demonstrated improvement via profiling or a reasonable
benchmark) then I think investigating doing it when that CPU type is
selected in 'make config' is well worth doing.
I also think squashing some of the gcc version and flag dependancies
is worth doing as a longer-term project, but that's a totally
different can of worms. I'm not even sure it's possible to do it with
gcc _yet_. It seems to me though, there's been quite a bit of progress
in getting away from some of the compiler-specific issues.
--
Matt Gauthier <[EMAIL PROTECTED]>
------------------------------
From: Pete Becker <[EMAIL PROTECTED]>
Crossposted-To: comp.lang.c++,gnu.misc.discuss
Subject: Re: Compiling C++ programs with GCC --> no GPL license implications
Date: Tue, 12 Dec 2000 10:49:44 -0800
Ken Arromdee wrote:
>
> In article <[EMAIL PROTECTED]>,
> Pete Becker <[EMAIL PROTECTED]> wrote:
> >> Tisdale is a known troll. Save your keystrokes.
> >Bad idea. He happens to be right. Rights under copyright law are decided
> >by statues and by courts, not by public announcements.
>
> *Giving up* a right can be done by public announcements.
Perhaps, if the person doing it has authority to bind everyone who is in
a position to enforce that right. Before you risk your multi-million
dollar product on someone waiving their rights under the copyright law,
you better have something more than an assurance you saw in an e-mail
somewhere.
Legal opinions that you get through newsgroups aren't worth the paper
they're printed
on.
--
Pete Becker
Dinkumware, Ltd. (http://www.dinkumware.com)
Contributing Editor, C/C++ Users Journal (http://www.cuj.com)
------------------------------
From: "Marty" <[EMAIL PROTECTED]>
Subject: source code of mkfs.minix
Date: Wed, 13 Dec 2000 01:01:50 +0800
How can I find the source code of /sbin/mkfs.minix ?
------------------------------
From: "Mike Austin" <[EMAIL PROTECTED]>
Subject: Device Driver
Date: Tue, 12 Dec 2000 12:27:54 -0500
I'm developing a device Driver under RedHat 6.2/7.0 for a PCI board using a
V3 chip. The driver works perfectly fine when one card is in the system, I
only see a problem with multiple cards. In the 2 card situation, the first
card does everything I want it to (memory read/writes, interrupts, DMA
transfers). The second card, however, fails on the DMA transfers. The V3
reports that the DMA is finished (DMAlength drops to 0), yet the dma in
action bit is still asserted, and the interrupt pending bit is not set. Any
memory writes to the second card after this results in a system crash. Do
you know of any problems with this configuration as far as either hardware
or Linux is concerned? Is there a software fix?
Thank you!
Michael Austin
------------------------------
From: Alex Graf <[EMAIL PROTECTED]>
Subject: Re: changing BASH's path searching
Date: Tue, 12 Dec 2000 11:42:02 -0800
Reply-To: [EMAIL PROTECTED]
Josef Moellers wrote:
>
> There's more to rpm and friends than just injecting files here and there
> into a directory tree, checking dependencies for one and the most
> important thing: documenting what's installed.
> Rather than doing a "find / -name command -print" (and going to lunch)
> to see if a command is installed ("Rats, did I install the help files
> with them?" followed by another "find" and another lunch break),
> searching the rpm database to decide whether the package is installed
> and if it's the right version is a lot easier.
> Also, removing a package is a lot easier: No leftovers (err ... not from
> lunch), no broken dependencies, documented removal.
But if every package had its own directory, then you don't need to
search with find, or even look it up in a package database -- all you
need to do is check if its directory is there or not.
And if *every* file for a package were contained in a single directory,
docs, libs, global configs, executables & all, then you can be
absolutely sure there are no leftovers by wiping out the whole
directory: rm -Rf <pkg-dir>
In older debians that I've used, dpkg --purge would often leave files &
dirs behind.
As many have pointed out to me, my idea stinks in a networked
environment with shared directory trees, multiuser, etc. I completly
agree with all of them. I probably should have clarified that I am
interested in scaled-down linux for small isolated single-user machines
with no knowledgeable sysadmin, only users who want to install new
software (maybe downloaded from the 'net, maybe from a store). In any
other role, I agree with everyone that apt is the best way to go.
> > on keeping around crufty old /usr, /var, /etc. Trying to be old-school
> > I guess.
>
> Compared to one \WINDOWS\SYSTEM directory, the scheme with /usr, /opt,
> /var, /etc, /lib sure sounds a little more structured.
I agree that my statement was too rash!
OTOH I never suggested that packages should throw files into a common
SYSTEM directory.
> No, I'd rather have a few centralized locations where I (and my shell)
> can look for commands. It saves a lot of directory accesses and speeds
> up command execution considerably. Symbolic links have been mentioned
OK, that's very true, there would be too many directory accesses and it
would be too slow. How about a $PATH search and *then* a subdirectory
search?
Alex
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.lang.c++,gnu.misc.discuss
Subject: Re: Compiling C++ programs with GCC --> no GPL license implications
Date: Tue, 12 Dec 2000 18:46:31 GMT
In article <[EMAIL PROTECTED]> Pete Becker <[EMAIL PROTECTED]>
writes:
<snip>
> Rights under copyright law are decided
> by statues and by courts, not by public announcements. Legal opinions
> that you get through newsgroups aren't worth the paper they're printed
> on.
>
> --
> Pete Becker
> Dinkumware, Ltd. (http://www.dinkumware.com)
> Contributing Editor, C/C++ Users Journal (http://www.cuj.com)
I would not call what I posted an authorative legal opinion but the
results of a research effort shared with others to reach their own
conclusions. I also hope people are not wasting paper on any newsgroup
posts ;-)
That said, information in newsgroups does have some value otherwise we
wouldn't be reading them so much.
Pete has a good point about the legalness of what appears in newsgroups.
It's difficult for the reader to discern whether something is true (and
thus would hold up in court) or not except through the thoroughness of
the argument itself, the trust they have in the others who post their
comment, and the trust they have in the system that shows those comments
actually came from who they thought they came from. These things can be
traced definitively (electronic fingerprints) but not by the average
reader. I believe this applies to email as well.
Yet the core of this discussion is correct and I believe FSF will stand
behind what I have communicated (as we all went to considerable trouble
to resolve this particular point).
It would be more definitive for the FSF to post their own synopsis of
this discussion on their website and/or include it as a text file with
future distributions of GCC. That should be more definitive as it is
somewhat more difficult for websites and the GCC distribution itself to
be faked. I'll query FSF directly to see if this discussion and it's
conclusions can be acknoledged on gnu.org somewhere.
On another note, it's my understanding that Dinkumware is trying to make
a business out of selling a version of libstdc++ that has clearer
license terms (as well as other potential features), and this situation
may make it less likely that people will buy their libstdc++ library
based on licensing concerns with the GCC libstdc++.
Regards,
Ralph T. Wayland
Sent via Deja.com
http://www.deja.com/
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.lang.c++,gnu.misc.discuss
Subject: Re: Compiling C++ programs with GCC --> no GPL license implications
Date: Tue, 12 Dec 2000 19:00:55 GMT
> Kaz Kylheku wrote:
<snip>
> > Tisdale is a known troll. Save your keystrokes.
<snip>
"Trolls" have their uses. I've known many. In this case it has sparked
useful discussion. But we do need to beware being dragged down into
their often labyrinthian caves ;-)
Regards,
Ralph T. Wayland
Sent via Deja.com
http://www.deja.com/
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: source code of mkfs.minix
Date: Tue, 12 Dec 2000 19:17:03 -0000
On Wed, 13 Dec 2000 01:01:50 +0800 Marty <[EMAIL PROTECTED]> wrote:
| How can I find the source code of /sbin/mkfs.minix ?
Much the same way as any other source code. Your distribution should
include it in the source code section. Look in the utilities files.
Or download it from a major site like metalab.unc.edu (again, in the
utilities sections).
--
=================================================================
| Phil Howard - KA9WGN | Dallas | http://linuxhomepage.com/ |
| [EMAIL PROTECTED] | Texas, USA | http://phil.ipal.org/ |
=================================================================
------------------------------
From: [EMAIL PROTECTED] (Dave Platt)
Crossposted-To: linux.dev.kernel
Subject: Re: slow asymmetric half-duplex communication (ADSL)
Date: Tue, 12 Dec 2000 19:19:14 -0000
>I recently got an ADSL connection (640kbaud down, 90kbaud up,
>PPPoE). Installation went well, but a simultaneous upload &
>download revealed that the download speed suffers big time
>when uploading takes place at the same time (total speed <
>120kbaud). I later learned that ADSL is indeed using a
>half-duplex protocol!
Industry-standard ADSL (the G.lite or similar) is not half-duplex, as
far as I know. As I understand it, it uses two non-overlapping sets
of frequencies - the uplink frequencies run from about 80 kHz to
somewhere around 200 kHz, and the downlink frequencies run from there
up to a bit above 100 MHz. The DMT (discrete multitone) carrier sets
should not interfere with one another.
> I could't believe it. Is there a way
>to tell the kernel to throttle an interface's outgoing packets
>to increase its overall performance? Packet schedulers come to
>mind.
I believe you've diagnosed the illness incorrectly, but (somewhat
paradoxically) are looking for the correct medicine.
I suspect that you aren't running into a full-duplex / half-duplex
problem. Instead, you're running into an ACK problem. The data being
downloaded is being sent via TCP, and this protocol requires that the
receiving system periodically send back an ACK packet which updates
the "receive window". When you're uploading at full speed, the ACK
packets for the download connection are being delayed - they're ending
up in the transmit queue (either the Ethernet card, or the ADSL modem)
behind the massive globs of data you're uploading. Because the ACKs
are delayed, the sending system throttles itself (believing that it's
overloading the downlink) according to the standard TCP sliding-window
rules and congestion-management rules.
So, what you probably want to do is to give the ACK packets priority
over the bulk data. You can very probably do this with the "traffic
shaping" features that are available in 2.4 (and as a beta-grade
experimental feature in the 2.2 kernels).
Take a look at /usr/src/linux/Documentation/networking/shaper.txt for
starters - I believe there are some more advanced HOWTO documents out
on the Web as well.
I haven't set up the shaper myself, so I can't really advise you in
detail on how to go about segregating the traffic. It may be
sufficient to set up two output queues - one for small packets (say, <
100 bytes or so) and one for all other packets, and give the
small-packet queue higher priority. This should let the ACKs, and
interactive-keystroke traffic leapfrog in front of the big
data-transfer packets, and should reduce or eliminate the TCP window
starvation.
You might also want to try tweaking the sources for the FTP client
program. You might be able to add in an ioctl() call so that it would
set the TCP class-of-service of the outgoing data connection to
"bulk". Many routers and network stacks will give such data lower
priority for transmission - you can definitely use the class-of-
service as a way of doing packet queueing management when running the
Linux traffic shaper.
--
Dave Platt [EMAIL PROTECTED]
Visit the Jade Warrior home page: http://www.radagast.org/jade-warrior/
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!
------------------------------
From: Pete Becker <[EMAIL PROTECTED]>
Crossposted-To: comp.lang.c++,gnu.misc.discuss
Subject: Re: Compiling C++ programs with GCC --> no GPL license implications
Date: Tue, 12 Dec 2000 14:25:00 -0800
[EMAIL PROTECTED] wrote:
>
> On another note, it's my understanding that Dinkumware is trying to make
> a business out of selling a version of libstdc++
We do not sell any version of libstdc++. We do sell our standard library
for use with gcc.
> that has clearer
> license terms (as well as other potential features), and this situation
> may make it less likely that people will buy their libstdc++ library
> based on licensing concerns with the GCC libstdc++.
Technical and quais-legal discussions should be based on facts, not on
innuendo. I make no apologies for urging people to get sound legal
advice before making business decisions that involve licensing issues.
That is simply common sense.
--
Pete Becker
Dinkumware, Ltd. (http://www.dinkumware.com)
Contributing Editor, C/C++ Users Journal (http://www.cuj.com)
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: imaginary complex value
Date: Tue, 12 Dec 2000 19:29:14 -0000
On Tue, 12 Dec 2000 14:40:37 GMT Dr. Unk <[EMAIL PROTECTED]> wrote:
| While compiling glibc-2.2 I get an error, since I have little
| programming experience (other than a few programming projects at
| school) I cannot fix this myself. I suspect I might have something to
| do with gcc, in the past I have come upon problems with "imaginary"
| numbers while bootstrap compiling both GCC and EGCS. I'm totaly
| confused and would like to know if there is a fix to this. Below is
| the error output given to me while attempting to compile the math part
| of glibc:
| In file included from ../include/bits/cmathcalls.h:1,
| from ../math/complex.h:81,
| from ../include/complex.h:1,
| from conj.c:21:
| ../math/bits/cmathcalls.h: In function `cimagf':
| ../math/bits/cmathcalls.h:139: Unable to access imaginary part of
| complex value
| in a hard register on this target
| make[2]: *** [/opt/tmp/glibc-2.2/build/math/conj.o] Error 1
| make[2]: Leaving directory `/opt/tmp/glibc-2.2/math'
| make[1]: *** [math/others] Error 2
| make[1]: Leaving directory `/opt/tmp/glibc-2.2'
| make: *** [all] Error 2
It may be the case that the cimag function is being called with an
object that is not, or cannot be, a complex number. Since it says
"hard register" it seems something wrong is being passed through.
This isn't really a systems issue. Why not provide source code to
the comp.os.linux.development.applications newsgroup and see if they
can give you any better info.
--
=================================================================
| Phil Howard - KA9WGN | Dallas | http://linuxhomepage.com/ |
| [EMAIL PROTECTED] | Texas, USA | http://phil.ipal.org/ |
=================================================================
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To:
alt.os.linux.mandrake,alt.os.linux.slackware,alt.uu.comp.os.linux.questions,comp.os.linux.misc,comp.os.linux.networking,comp.os.linux.hardware,comp.os.linux.setup,linux.dev.kernel
Subject: kernel questions
Date: Tue, 12 Dec 2000 20:05:10 GMT
Reply-To: [EMAIL PROTECTED]
I have a couple of questions.
What is a backport?
What patch goes with what kernel? i.e. Does patch-2.2.18 update kernel
2.2.17 or 2.2.18?
best regards,
charles
------------------------------
** 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 by posting to the
comp.os.linux.development.system newsgroup.
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
******************************
