Linux-Development-Sys Digest #653, Volume #7 Sun, 5 Mar 00 21:13:14 EST
Contents:
Re: Absolute failure of Linux dead ahead? ([EMAIL PROTECTED])
Re: Binary compatibility: what kind of crack are they smoking? (Colin Watson)
Re: Installing glibc (Please Help!) (Andreas Jaeger)
blocksize craziness in 2.2.14 ([EMAIL PROTECTED])
Seeking Linux developer for open source project ("Dean Roddey")
what is the relative overhead between ... ([EMAIL PROTECTED])
Re: GCC and EGCS, SuSE say they can co-exist, I don't see how (Lee Reynolds)
What are the reasons for using different compiler revisions? (Lee Reynolds)
Re: What are the reasons for using different compiler revisions? (Jeff Biviano)
Re: modem driver (Lee Reynolds)
Re: What are the reasons for using different compiler revisions? (Paul Kimoto)
Re: Seeking Linux developer for open source project (Christopher Browne)
Re: array of semaphores... ("mr mike (mike wingert)")
Re: Seeking Linux developer for open source project ("Dean Roddey")
----------------------------------------------------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.os.linux.advocacy
Subject: Re: Absolute failure of Linux dead ahead?
Reply-To: [EMAIL PROTECTED]
Date: Sun, 05 Mar 2000 16:14:43 GMT
[EMAIL PROTECTED] (Jon) writes:
>> Jon <[EMAIL PROTECTED]> wrote:
>> > (Wolfgang Weisselberg) wrote:
>> > > How many machines do *you* know that are in active use today
>> > > *and* were so 15,20,30 years ago?
>>
>> > 2 that I've worked with personally.
[...]
>One of the 2 machines is (yes, it's still in use) a 386DX40 with
>32MB RAM and an RLL drive on a 16MB cache card. I have 4
>machines in pieces at home that outpower that thing.
A 386DX40 is less than 10 years old. Not 15, not 20, not 30, and most
certainly not 38.
Bernie
--
You see things, and you say 'Why?' But I dream things that never were,
and say 'Why not?'
George Bernhard Shaw
Irish playwright, 1856-1950
------------------------------
From: [EMAIL PROTECTED] (Colin Watson)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.setup
Subject: Re: Binary compatibility: what kind of crack are they smoking?
Date: 5 Mar 2000 13:08:55 GMT
Donovan Rebbechi <[EMAIL PROTECTED]> wrote:
>On 25 Feb 2000 14:18:02 GMT, Kari Pahula wrote:
>>It is then distribution makers' responsibility to compile newest
>>versions compatible with distribution-specific libraries, if they
>>choose to do so. Either way, the user is always able to compile the
>>programs herself.
>
>This assumes that there exist no third party binary only apps. This in
>the long term is a dangerous assumption.
Actually, I suspect that in the short term it's a dangerous assumption,
but in the long term it's quite a safe one, at least if the current
open-source climate continues.
>BTW, it would be kind of cool if the build system were more failsafe
>( like the BSD ports ). BSD ports basically automatically download all
>the compile time dependencies, and install them ( trying multiple ftp
>sites in case one is down ) , and the software is verified using checksums
>that come with the distribution. This way, unattended builds are much simpler.
This is on its way in Debian, you'll be glad to hear; Build-Depends: are
not mandatory yet (I'm not even sure if they're mentioned in policy),
but the plan is eventually that you'll be able to build Debian source
packages like this.
--
Colin Watson [[EMAIL PROTECTED]]
"It's the one who won't be taken who cannot seem to give,
And the soul afraid of dying that never learns to live."
------------------------------
From: Andreas Jaeger <[EMAIL PROTECTED]>
Subject: Re: Installing glibc (Please Help!)
Date: 05 Mar 2000 19:19:49 +0100
>>>>> Martin Wackenhut writes:
> Hi everybody!
> I allready posted this message a few days ago but nobody answered. So
> please help me this time.
> I need to know how to Install the glibc "by hand" this means:
> Copiing every single file to the exact location and creating the
> specific symlinks.
> Has anybody an idea or a good documentation telling me all this.
> That would be great.
The glibc comes with enough documention already. Please read the FAQ,
INSTALL and README files.
Andreas
--
Andreas Jaeger
SuSE Labs [EMAIL PROTECTED]
private [EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED]
Subject: blocksize craziness in 2.2.14
Date: Sun, 05 Mar 2000 19:49:05 GMT
kernel is version 2.2.14
mke2fs is version 1.15
If I wipe a partition to zeros with bs=1024:
dd if=/dev/zero of=/dev/hda5 bs=1024
then format a filesystem with -b 4096:
mke2fs -b 4096 -i 8192 /dev/hda5
then mount it:
mount /dev/hda5 /tmp
then unmount it:
umount /tmp
then try to wipe it to zeros with bs=1024 again:
dd if=/dev/zero of=/dev/hda5 bs=1024
I get this error:
dd: /dev/hda5: Input/output error
21+0 records in
20+0 records out
then try to wipe it to zeroes with bs=4096:
dd if=/dev/zero of=/dev/hda5 bs=4096
there are no errors.
If I perform the above w/o doing the mount/umount sequence, the errors
do not occur.
Also the following command sequence (first 4 same as above):
dd if=/dev/zero of=/dev/hda5 bs=1024
mke2fs -b 4096 -i 8192 /dev/hda5
mount /dev/hda5 /tmp
umount /tmp
mke2fs -b 1024 -i 8192 /dev/hda5
I get a massive number of errors like:
Writing inode tables: 0/40Warning: could not write 8 blocks in inode table starting
at 22: Attempt to write block from filesystem resulted in short write
Warning: could not write 8 blocks in inode table starting at 38: Attempt to write
block from filesystem resulted in short write
Warning: could not write 8 blocks in inode table starting at 54: Attempt to write
block from filesystem resulted in short write
Warning: could not write 8 blocks in inode table starting at 70: Attempt to write
block from filesystem resulted in short write
Warning: could not write 8 blocks in inode table starting at 86: Attempt to write
block from filesystem resulted in short write
Warning: could not write 8 blocks in inode table starting at 102: Attempt to write
block from filesystem resulted in short write
Warning: could not write 8 blocks in inode table starting at 118: Attempt to write
block from filesystem resulted in short write
[this repeats many times with increasing numbers after "starting at"]
Again, the problem does NOT happen if I skip the mount/umount step.
While I cannot totally exclude a hardware problem in the drive itself,
Further tests have found that if I never do the mount/umount I never
get any errors writing to the drive.
What I suspect may be happening is that the blocksize setting may be
affecting the way the disk blocks are cached, and when the setting is
changed the cache is not handled appropriately (perhaps changing the
setting in each cached block, or flushing the entire cache for that
device). Executing "sync" between each command doesn't change anything.
But if the cache is written OK, remains in memory as the wrong blocksize
(is it associating blocksize in cache?), IWSTM, that could interfere
with subsequent I/O at a different blocksize.
Possibly the hardware is getting a blocksize setting and it is not being
reset properly? IWSTM that if the device had such a setting (does it?)
and the kernel sets it, such as from the mount, it should know to set it
back for any subsequent I/O, or reblock any I/O appropriately. IOW, if
the device has a blocksize setting the kernel driver for the device should
properly track and control it (but I don't know if that is a meaningful
scenario here, just guessing).
Does anyone know exactly what _is_ going on here?
Any ideas of a test sequence I could try to narrow down the problem?
The system I am doing this on is a new install, so early suggestions
can be destructive to data since I can just re-install w/o really
losing anything but time (and I've already played with it for 3 hours).
--
| Phil Howard - KA9WGN | for headlines that | Just say no to absurd patents |
| [EMAIL PROTECTED] | really matter: | Boycott Amazon.Com (AMZN) |
| Dallas - Texas - USA | linuxhomepage.com | Shop http://bn.com/ instead |
------------------------------
From: "Dean Roddey" <[EMAIL PROTECTED]>
Subject: Seeking Linux developer for open source project
Date: Sun, 5 Mar 2000 12:08:10 -0800
Reply-To: "Dean Roddey" <[EMAIL PROTECTED]>
I'm the author of a large, portable, open source product called the CIDLib
C++ Frameworks. This product currently has about 350 classes, and has been
in existence for about 8 years at this time. It recently went open source.
Main development is on NT/VC++, but its very portable and just requires a
driver layer for portability. There is currently a driver for Linux, but the
person who wrote it wants to move on, and I really need someone to pick up
this project and perfect it.
You would be providing an important service to the Linux and open source
communities, since CIDLib provides for 'as is' portability of code between
Linux and NT. So developers can provide important services to the Linux
community while still retaining access to the lucrative NT server market.
If you might be intested, you can check out the product at
www.charmedquark.com where there is a complete tutorial and class reference
on line. CIDLib is also now available on SourceForge at
http://sourceforge.net/project/?group_id=2390 which provides bug tracking,
mailing lists, task lists, FTP, and a CVS source repository to support
distributed development.
Please, if you are experienced in Linux and C++, and would be interested in
working on the Linux driver (or extending it to other Unix platforms as
well), or implementing some other high level functionality on top of the
CIDLib system, please contact me.
==========================
Dean Roddey
The CIDLib Class Libraries
Charmed Quark Software
[EMAIL PROTECTED]
http://www.charmedquark.com
"Give me immortality, or give me death"
------------------------------
From: [EMAIL PROTECTED]
Subject: what is the relative overhead between ...
Date: Sun, 05 Mar 2000 20:45:11 GMT
What is the relative overhead between ...
1. A process that dlopen()'s an executeable and directly calls a desired
function within.
2. A process that (w/o forking) calls execv() to run an executeable in
which main() will call the desired function within.
--
| Phil Howard - KA9WGN | for headlines that | Just say no to absurd patents |
| [EMAIL PROTECTED] | really matter: | Boycott Amazon.Com (AMZN) |
| Dallas - Texas - USA | linuxhomepage.com | Shop http://bn.com/ instead |
------------------------------
From: Lee Reynolds <[EMAIL PROTECTED]>
Subject: Re: GCC and EGCS, SuSE say they can co-exist, I don't see how
Date: Sun, 05 Mar 2000 19:41:02 -0700
> I want 2.7.2.3 for kernel builds, but EGCS for software development.
> I fail to see how to do this and have GCC somehow run one then the
> other.
Why use different compilers? I downloaded and compiled gcc 2.95.2 and I
use it for everything. I've never had any problems and I compile all
kinds of things, including the 2.3.x kernels. I'm not trying to tell you
that you are wrong in your opinion, I'm just trying to find out what the
reason is for it.
Lee
------------------------------
From: Lee Reynolds <[EMAIL PROTECTED]>
Subject: What are the reasons for using different compiler revisions?
Date: Sun, 05 Mar 2000 19:45:42 -0700
I've noticed that at least one person here likes to use GCC 2.7.2 to
compile kernels with, but newer revisions such as EGCS or GCC 2.95.2 for
application development. What is the reason for this? Does the earlier
compiler generate better code? Or are there bugs in the newer compilers
that haven't quite gotten worked out yet?
Lee
------------------------------
From: Jeff Biviano <[EMAIL PROTECTED]>
Subject: Re: What are the reasons for using different compiler revisions?
Date: Sun, 05 Mar 2000 22:38:42 GMT
Lee Reynolds wrote:
> I've noticed that at least one person here likes to use GCC 2.7.2 to
> compile kernels with, but newer revisions such as EGCS or GCC 2.95.2 for
> application development. What is the reason for this? Does the earlier
> compiler generate better code? Or are there bugs in the newer compilers
> that haven't quite gotten worked out yet?
>
> Lee
Some versions of the Kernel (Especially 2.0.x series) take advantage of
quirks in the 2.7.2 compiler. These quirks are not present in more recent
compilers. The new compilers are preferred since much improvments have
been made to produce smaller more efficient code, but if they cannot
compile the Kernel the easy route is to use the 2.7.2 compiler.
- Jeff
------------------------------
From: Lee Reynolds <[EMAIL PROTECTED]>
Subject: Re: modem driver
Date: Sun, 05 Mar 2000 20:16:16 -0700
Terry Williford wrote:
> I would like to set up linux on my system today. However, I need a device
> driver for my modem. It is a Best Data 56Fw pci modem. Any suggestions?
>
> Thanks in advance....
Replace it. It is a winmodem, which means it will not run under linux unless
you are willing to write the software layer it depends on to function.
Real modems do all their own work. A computer talks to them the way it would
any other device hooked up to a serial port. They will work with any computer
that is fast enough to talk to them and has the right kind of serial port.
Winmodems on the other hand don't do their own work. They rely on special
software in the computer to translate the signals coming in off the phone line
into data that the computer can use. Linux doesn't have this software layer
because the companies which make these modems have not sat down and created
it.
All you can do now is buy a new modem, preferably an external one. Externals
only cost about $10 to $15 more than a _real_ internal modem. They don't eat
up an ISA slot, and you can hook them up to another system very easily if you
want. Also, the blinky lights can give you important information as to how
your connection to the internet is behaving. If you're looking for brand
names or models to buy, Zoom makes good external modems. We've had one here
for almost two years on our IP masquerading gateway and it's performed
admirably.
Lee
------------------------------
From: [EMAIL PROTECTED] (Paul Kimoto)
Subject: Re: What are the reasons for using different compiler revisions?
Date: 5 Mar 2000 17:52:48 -0500
Reply-To: [EMAIL PROTECTED]
In article <[EMAIL PROTECTED]>, Lee Reynolds wrote:
> I've noticed that at least one person here likes to use GCC 2.7.2 to
> compile kernels with, but newer revisions such as EGCS or GCC 2.95.2 for
> application development. What is the reason for this? Does the earlier
> compiler generate better code?
Kernel code may rely on undocumented gcc behavior, i.e., behavior that the
gcc maintainers may change from time to time. Hence, when Alan Cox was
maintaining 2.0.*, gcc-2.7.* was required as the compiler. Now, as 2.2.*
maintainer, he says that gcc-2.95.* is not guaranteed to give the right
results, although for many people it has been working well.
--
Paul Kimoto <[EMAIL PROTECTED]>
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Seeking Linux developer for open source project
Reply-To: [EMAIL PROTECTED]
Date: Sun, 05 Mar 2000 22:53:27 GMT
Centuries ago, Nostradamus foresaw a time when Dean Roddey would say:
>You would be providing an important service to the Linux and open source
>communities, since CIDLib provides for 'as is' portability of code between
>Linux and NT. So developers can provide important services to the Linux
>community while still retaining access to the lucrative NT server market.
WxWindows makes similar claims about being valuable in the GUI realm
as a framework usable with both Windows and UNIX, with GTK and Win32
and Motif. Unfortunately, it's not a bridge that allows you to take
Win32 apps and redeploy them on UNIX; it's a bridge where if you write
applications in The WxWindows Way, you can deploy on a couple of
platforms.
I don't want to be overly negative, but wouldn't it be more accurate
to say that:
"*If* this work gets done, and *If* people write a bunch of
applications that then use this framework, there is significant
benefit to the Linux community"?
--
Microsoft: The Scientology of Computing
[EMAIL PROTECTED] - - <http://www.ntlug.org/~cbbrowne/lsf.html>
------------------------------
From: "mr mike (mike wingert)" <[EMAIL PROTECTED]>
Crossposted-To: comp.programming.threads,comp.os.linux.development.apps
Subject: Re: array of semaphores...
Date: Sun, 05 Mar 2000 18:30:43 -0500
The key to getting help with profs is to NAG NAG NAG. works every time =-)
>
> I believe, that this is not true in my case. I wish, I had prof. who could
> answer all of my questions,
> but the prof.s here don't care about students. They just lecture and leave,
> and TA's do marking and explanations.
> The TA's themselves may not even know the course material. :(
--
================================================================
The Hubble Space Telescope exists to find signs of cheese in other
celestial bodies, be they star system or stars themselves. The
grand lunar cheese expeditions were a phenominal disaster due to
the total lack of cheese on the moon, as had been previously believed.
Subsequent missions were sent after the initial to see if maybe they
were just looking in the wrong place.
-anonymous
------------------------------
From: "Dean Roddey" <[EMAIL PROTECTED]>
Subject: Re: Seeking Linux developer for open source project
Date: Sun, 5 Mar 2000 17:45:53 -0800
Reply-To: "Dean Roddey" <[EMAIL PROTECTED]>
> WxWindows makes similar claims about being valuable in the GUI realm
> as a framework usable with both Windows and UNIX, with GTK and Win32
> and Motif. Unfortunately, it's not a bridge that allows you to take
> Win32 apps and redeploy them on UNIX; it's a bridge where if you write
> applications in The WxWindows Way, you can deploy on a couple of
> platforms.
>
> I don't want to be overly negative, but wouldn't it be more accurate
> to say that:
> "*If* this work gets done, and *If* people write a bunch of
> applications that then use this framework, there is significant
> benefit to the Linux community"?
Well, I would have assumed that this went without saying. Of course, you
have to use a framework in order for the framework to present you any
benefits :-) I said it provides "'as is' portability of code between Linux
and NT", not that you could take Win32 API applications and run them on
Linux. Even if it did do that, it wouldn't be remotely doing what I think
that a framework should do, which is (in addition to portability) provide a
much, much more high level, safe, and powerful programming environment. And
CIDLib certainly does that.
As for "if the work gets done", the work *is* 'done', to the degree that any
piece of software is ever 'done.' The current 1.1.0 release has a working
Linux driver in it. And all of the samples and utilities that ship work
absolutely as is, with zero conditional code. CIDLib is very abstracted from
the host platform, and gets its portability by keeping *all* system and
runtime headers out of client code and out of its own code, except in a
small platform/environment driver. I just need to find someone willing to
continue the driver Linux work and to make it better, faster, smaller,
etc...
I'm currently developing a net based bug tracking and source control program
based on CIDLib. Its more to prove the power of the libraries than because
I'm really interesting in that type of application (I'm a plumbing kind of
guy and and user apps don't push my buttons as much.) Other than the text
diff'er, which I'll write as a standard reusable CIDLib tool of course,
basically everything is there and ready to rock-n-roll. The small amount of
code required, and the easy portability (basically a recompile) will (I
hope) prove the power of CIDLib and hopefully get it some more attention.
And yes, it will only be useful if people use it, which is why I'm writing
this, to try to get people to use it and to help me bring this gift to the
people, so to speak.
==========================
Dean Roddey
The CIDLib Class Libraries
Charmed Quark Software
[EMAIL PROTECTED]
http://www.charmedquark.com
"Give me immortality, or give me death"
------------------------------
** 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
******************************
