Linux-Misc Digest #215, Volume #21 Thu, 29 Jul 99 20:13:10 EDT
Contents:
Killing unkillable files ("M. Leo Cooper")
Re: math.h problem (Dan Pop)
Re: What I think of linux. (uRBaN_WeaSeL)
Re: windows dll vs. linux libraries (mlw)
Re: Linux has finally crashed (Andreas Dilger)
LILO and NT40 (Matt Menze)
Re: Linux has finally crashed (mlw)
Re: Need script to convert filenames from UPPER to lower (Christopher Browne)
Re: Is RPM unique to Linux ? (Christopher Browne)
Re: Is Linux A Memory Hogging OS? (Christopher Browne)
Re: Is RPM unique to Linux ? (Christopher Browne)
Re: WWW: Practical Enterprise Linux/Open Source Site Launched (Christopher Browne)
Re: Scripting Question (Christopher Browne)
Re: GL/Mesa with X11 (Christopher Browne)
----------------------------------------------------------------------------
From: "M. Leo Cooper" <[EMAIL PROTECTED]>
Subject: Killing unkillable files
Date: Thu, 29 Jul 1999 15:11:53 -0700
Does anyone have any insight as to how to kill dormant and zombie
processes, the ones that a kill -9 or even a kill -15 won't terminate?
This would be very helpful, as certain running processes will prevent a file
system umount on shutdown (kernel 2.2.9, RH6.0).
Thanks.
------------------------------
From: [EMAIL PROTECTED] (Dan Pop)
Crossposted-To: comp.lang.c,gnu.gcc.help
Subject: Re: math.h problem
Date: 29 Jul 99 22:22:38 GMT
In <[EMAIL PROTECTED]> Ben Pfaff <[EMAIL PROTECTED]> writes:
>[EMAIL PROTECTED] (Dan Pop) writes:
>
> In <[EMAIL PROTECTED]> Ben Pfaff <[EMAIL PROTECTED]> writes:
>
> >[EMAIL PROTECTED] (Dan Pop) writes:
> >
> > The problem can be fixed without breaking the -lm kludge or *any*
> > Makefile rules relying on it: put all the math stuff in *both* libc and
> > libm.
> >
> >Wouldn't it be less wasteful to put math in libc and provide an empty
> >libm?
>
> This will break some esoteric linking procedures which need math functions
> but do not include libc, either explicitly or implicitly.
>
>Doesn't the existence of such linking procedures imply that there is
>good separation of libm from libc, and that in fact they should not be
>merged?
Merging them, as I proposed above, would not break those procedures.
So, what exactly is your point? I could understand your objection if I
suggested putting all the stuff from libc into libm, which I didn't.
Dan
--
Dan Pop
CERN, IT Division
Email: [EMAIL PROTECTED]
Mail: CERN - IT, Bat. 31 1-014, CH-1211 Geneve 23, Switzerland
------------------------------
From: uRBaN_WeaSeL <[EMAIL PROTECTED]>
Crossposted-To: alt.linux,alt.linux.sux,alt.os.linux,comp.os.linux.advocacy
Subject: Re: What I think of linux.
Date: Thu, 29 Jul 1999 18:18:10 -0400
Jay Fink wrote:
>
> Crap I totally mucked this up:
>
> >
> Linux inherently has tons more power than Windows and more stability -
> but - the end user interface is rough for Windows or MAC users.
I just came from Win95 and decided to use Linux exclusively. Actually, I
didn't have too much trouble using XFree86 and WindowMaker. I admit it DOES
take some getting used to but I find it quite usable. And FAR more stable than
Windoze ever was!
> The "new fangled" is a reference to the Interface, not the Core OS.
> Most end users are more concerned with task oriented eye candy.
Gotta admit, me too. Even though I can (barely) navigate at the command
prompt, and maybe write a very simple shell script (if one dare call it even
that), I find myself 99% of the time in X-Windows.
------------------------------
From: mlw <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps
Subject: Re: windows dll vs. linux libraries
Date: Thu, 29 Jul 1999 23:34:01 +0000
Ulrich Weigand wrote:
>
> mlw <[EMAIL PROTECTED]> writes:
>
> >Also, I have not said the shared library paradigm that Linux has is bad,
> >I simply said it is not a good model for distributing binary objects to
> >a heterogenius installed base.
>
> >I have called many times for the concept of a shared module in addition
> >to shared libraries. Just think about being able to distribute a .so
> >file that can be used on any version of Linux, by any application no
> >mater what c library it is linked to.
>
> Well, using explicit dlopen() calls without the RTLD_GLOBAL flag and/or
> using versioned symbols you should be able to achieve an equivalent
> effect under Linux ...
>
> But if you really want this kind of independent objects, wouldn't a
> real object model be more appropriate? Under Windows, you'd usually
> use COM (or something on top of COM, like OLE or ActiveX). The corresponding
> mechanism under Unix would be CORBA (which is more like DCOM, but that
> shouldn't matter) ...
As I understand it DCOM and CORBA are more IPC mechanisms, and not
linking methodolgies.
--
Mohawk Software
Windows 95, Windows NT, UNIX, Linux. Applications, drivers, support.
Visit http://www.mohawksoft.com
------------------------------
From: [EMAIL PROTECTED] (Andreas Dilger)
Crossposted-To: comp.os.linux.development.apps
Subject: Re: Linux has finally crashed
Date: 29 Jul 1999 23:40:56 GMT
In article <[EMAIL PROTECTED]>,
David L. Johnson <[EMAIL PROTECTED]> wrote:
>My experience with AIX is (fortunately)�behind me, but the filesystem was not
>as reliable as linux. My AIX box would never re-boot after a power failure
>(which happens rather frequently at my U), and more than once files were
>corrupted.
I don't think that AIX JFS guarantees that the data in a file is
correct after an outage - only that the filesystem metadata is
consistent. If you had to guarantee the data, then your disk I/O speed
would be reduced by 50%, because you would have to write each piece of
data to the disk twice - once in the journal, and once on the data
area. An option under AIX is if you mirror your drives, you can select
that the copies are written one at a time, so that if the system
crashes you are guaranteed that one of the copies is OK. Note that
Oracle also keeps a transaction log of every bit of data written so it
can recover from a crash.
JFS does speed up a boot/fsck, since it only takes 30s to fsck a JFS
filesystem no matter what the size, unlike ext2. In my experience with
AIX, I've rarely, if ever, had a problem with filesystems at boot.
Cheers, Andreas
--
Andreas Dilger University of Calgary \"If a man ate a pound of pasta and
Micronet Research Group \ a pound of antipasto, would they
Dept of Electrical & Computer Engineering \ cancel out, leaving him still
http://www-mddsp.enel.ucalgary.ca/People/adilger/ hungry?" -- Dogbert
------------------------------
From: Matt Menze <[EMAIL PROTECTED]>
Subject: LILO and NT40
Date: Thu, 29 Jul 1999 16:46:01 -0700
I have a computer with NT4.0 on one hard drive and RH6.0 on another. I
was trying to set it up to use LILO so I could choose either linux or
NT. I installed the boot sector on the Master Boot Record (hda)
thinking that it would work. But now I cannot get NT to boot. I set up
LILO in linuxconf but when it tries to boot from my NT harddrive it
restarts LILO. I think that I may have overwritten the NT boot sector.
Is there anyway to create a startup disk from another NT workstation and
restore my NT boot sector?
------------------------------
From: mlw <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.x,comp.os.linux.development.apps
Subject: Re: Linux has finally crashed
Date: Thu, 29 Jul 1999 23:31:01 +0000
Graffiti wrote:
>
> In article <[EMAIL PROTECTED]>,
> mlw <[EMAIL PROTECTED]> wrote:
> [snip]
> >Stop right there. What I said is what I meant. Netscape locks up the
> >XServer and every things else. I have other machines here from which I
> >could telnet, but the system would not even ping.
>
> Then it's not a Netscape crash. It's a bug in the X server that Netscape
> happens to trigger. Blame the X server.
>
> -- DN
My post was:
"The only time I ever see this is a due to a flakey X server and using
Netscape. It has happened to me four times in three years."
I did blame the X server. Jeez.
--
Mohawk Software
Windows 95, Windows NT, UNIX, Linux. Applications, drivers, support.
Visit http://www.mohawksoft.com
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Need script to convert filenames from UPPER to lower
Reply-To: [EMAIL PROTECTED]
Date: Fri, 30 Jul 1999 00:05:00 GMT
On Thu, 29 Jul 1999 11:22:35 +0100, Jon Skeet <[EMAIL PROTECTED]> wrote:
>[EMAIL PROTECTED] wrote:
>
>> As I read the problem, we have a lot of files to rename, not a
>> file whose contents need translation.
>>
>> find . -type f -print | while read name
>> do
>> echo $name
>> mv $name `echo $name | tr A-Z a-z`
>> done
>>
>> That is, starting in the current directory, find all the ordinary files
>> and rename them with lowercase letters. Leave the directory names
>> (and pipes and device nodes...) alone.
>>
>> We assume, for simplicity, there aren't any lowercase-named files
>> lying around that would get clobbered by their similarly named
>> neighbors.
>>
>> Beware the tr(1) command. Christopher gave it an argument that only
>> GNU tr will understand. I gave an arg that GNU and BSD tar will know.
>> The System V tr prefers '[A-Z]' with brackets. The quotes would
>> prevent your shell from interpreting the brackets.
>
>Just as an aside, if you only want to do this once and don't want to
>bother with a full script, you can do it all in a single command:
>
>for i in `find . -type f`; do j=`echo $i | tr A-Z a-z`; mv $i $j; echo
>$i; done
>
>You certainly wouldn't want to type this very often though :)
This was indeed rather nearer to the solution than what I wrote; I
misread and thought that the issue was to find the filenames and
translate them, thinking that the filenames were actually in a file.
The way I'd do it if all the files were in one directory would be
(Korn shell):
for u in `ls`; do
l=`echo $u | tr '[:upper:]' '[:lower:]'`
mv $u $l
print "Moving $u to $l"
done
I'd consider it rather safer to have this build a script, thus:
touch mvscript
for u in `ls`; do
l=`echo $u | tr '[:upper:]' '[:lower:]'`
print `mv $u $l` >> mvscript
done
and then examine mvscript to make sure it's right before running it.
It is *incorrect* to say that the arguments were only usable by GNU
tr.
I was not at a Linux box, nor using GNU tr when I wrote it up. I
tested it on Solaris 5.5.1.
--
Next year in L5.
[EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/lsf.html>
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Is RPM unique to Linux ?
Reply-To: [EMAIL PROTECTED]
Date: Fri, 30 Jul 1999 00:05:03 GMT
On Thu, 29 Jul 1999 16:32:09 -0500, Matt Cole <[EMAIL PROTECTED]> wrote:
>> A newbie question.
>> I'm just wondering : Is RPM used only on Linux ?
>> Can RPM be used on FreeBSD ? and if not, why not ? Why aren't
>> applications packaged as RPM on other Unix platforms ?
>
>My understanding (and I could very well be wrong) is that RPMs are
>specific to RedHat. e.g., I use Slackware, and I can't use RPMs on my
>system...if I'm wrong about this, someone set me straight, please.
Your understanding is decidedly incorrect, particularly in that there
are two other notable Linux distributions that use RPMs, namely SuSE
and Caldera.
It is even possible to install and use RPM on a Slackware system,
albeit with the problem that packages designed with specific
provisions for the peculiarities of a particular distribution's
peculiarities, packages may not be *usefully* installable.
RPM can run on many other OSes, but as it is not generally the
"native" format for installing packages (the SYSV "pkg" system is more
widely used), it is often not terribly useful elsewhere.
--
"Computers in the future may weigh no more than 1.5 tons". -- POPULAR
MECHANICS magazine forecasting the "relentless march of science" 1955
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/lsf.html>
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: comp.os.linux.advocacy
Subject: Re: Is Linux A Memory Hogging OS?
Reply-To: [EMAIL PROTECTED]
Date: Fri, 30 Jul 1999 00:05:14 GMT
On Thu, 29 Jul 1999 13:59:36 -0400, Youngert <[EMAIL PROTECTED]> wrote:
>Christopher B. Browne <[EMAIL PROTECTED]> wrote in message
>news:[EMAIL PROTECTED]...
>> On Thu, 29 Jul 1999 12:53:51 -0400, Youngert
>> <[EMAIL PROTECTED]> posted:
>> >I have an AMD K6-2 400MHz, 128M RAM running SuSE-6.1 with Linux-2.2.10
>> >kernel. The computer's setup is a basic one with X11 + KDE. Everytime I
>> >compile the kernel, the system starts swapping at some point and never
>> >releases the memory even after finishing the kernel compilation. Is
>there a
>> >way to force the kernel to release the un-used memory?
>>
>> That sounds odd; with 128MB of RAM, there should be no need for it to
>> swap when merely compiling the kernel.
>>
>> That being said, ask yourself what "unused memory" the kernel should
>> be releasing.
>>
>> I *presume* that what you're referring to is the bits of memory that got
>> swapped out; you are expecting that eventually that the swapped-out
>> stuff will eventually, for some reason, get pulled back in.
>>
>> Don't expect that to happen. If the stuff that got pushed to swap
>> never gets referenced again, which might well be the case for (for
>> instance) the initialization code for XFree86, there's no value to
>> pulling it back into memory.
>> ...
>> In short, I think you're misinterpreting a feature as a bug.
>
>You are right. What my concern was more towards the swapped out data that
>never gets referenced and just stays in the Swap. From your explanation
>above, the kernel should release the XFree86 initialization code should
>there be of no value to pulling it back into the memory. Would not it make
>more sense?
This begs the question:
How does the system know that it is safe to throw away the code?
The initialization code isn't marked as being 'OK to throw away,' so
there's no way to say "This never need be loaded again.
Furthermore, in the case of XFree86, it includes code that *could* be
reinvoked if you changed modes, e.g. by hitting C-M-+ or C-M--, or if
you did other such things.
The Upcoming New Architecture in XFree86 4.0 splits it up into modules
that can be loaded/unloaded; in that situation, it may prove practical
to start things up, load in a module to do initialization, and then
unload the module.
That still doesn't do you anything more useful than having some bits
of data not sit in swap space.
It may offend your sensibilities, but it's doing you no damage,
performance-wise. Indeed, it downright *helps* you, performance-wise,
when you consider that by having disused code sit in swap-space, this
allows disk caches to be that much larger...
--
Rules of the Evil Overlord #60. "If I learn that a callow youth has
begun a quest to destroy me, I will slay him while he is still a
callow youth instead of waiting for him to mature."
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/lsf.html>
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Is RPM unique to Linux ?
Reply-To: [EMAIL PROTECTED]
Date: Fri, 30 Jul 1999 00:05:32 GMT
On 29 Jul 1999 14:51:15 GMT, John Robson <[EMAIL PROTECTED]>
wrote:
>A newbie question.
>I'm just wondering : Is RPM used only on Linux ?
>Can RPM be used on FreeBSD ? and if not, why not ? Why aren't
>applications packaged as RPM on other Unix platforms ?
RPM is known to run on such platforms as Solaris, HP-UX, SCO
OpenServer, Digital UNIX, SunOS, AIX, amongst others.
That suggests that it is pretty likely that it can run on the BSDs.
Users of some of those other platforms are a mite skeptical that they
*want* to use RPM to manage their software; there be Great Flame Wars
in that direction.
There is some controversy as to whether RPM represents the Best
Possible Tool for the purpose of managing software. I'd suggest that
the jury's still out on this.
- The Debian folk would legitimately complain that RPM doesn't provide
much in the way of tools to validate that packages install their
software in appropriate places.
- The *BSD folk feel it preferable to compile from source, using their
Ports system.
Both of thse groups Have A Point.
--
If I could put Klein in a bottle...
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/linuxdistributions.html>
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: WWW: Practical Enterprise Linux/Open Source Site Launched
Reply-To: [EMAIL PROTECTED]
Date: Fri, 30 Jul 1999 00:06:08 GMT
On Wed, 28 Jul 1999 08:22:59 -0700, Ernie DeVries
<[EMAIL PROTECTED]> wrote:
>> "The site is not meant to be a 'slashdot'.
>
>The site looks like it has some good content, but if you don't mean
>to be a 'slashdot' then why does it look like you just copied their
>pages & changed the logo at the top?
It uses the same software, as does Bruce Perens' new site, which
certainly results in similarity in presentation.
The point of the comment is that it is not intended to attract the
same degree of blithering idiots aka ``Anonymous Cowards.''
I suspect that some simple code changes could make it look a lot less
like Slashdot without changing functionality much; 'twould be wise to
do so...
--
QT adds to a Linux distribution a level of licencing complexity that
nullifies one of the major virtues of Linux: no licencing complexity.
-- <[EMAIL PROTECTED]>
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/lsf.html>
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: msn.computingcentral.os.linux
Subject: Re: Scripting Question
Reply-To: [EMAIL PROTECTED]
Date: Fri, 30 Jul 1999 00:06:08 GMT
On Wed, 28 Jul 1999 17:48:55 -0700, Jeff Grossman <[EMAIL PROTECTED]> wrote:
>Hello,
>
>I have a scripting question. Not sure if this would be done in Perl or
>something else. But here is what I need to do:
>
>1. I need to keep the last 5 versions of a file. I was thinking of just
>renaming the files, i.e. prev1, prev2, prev3, prev4, etc.
>2. So, I need to delete the last one, and rename all of the ones down one
>number.
The following Korn Shell script will do the trick; it would look very
similar in Bourne shell or C shell. Perl wouldn't give much
advantage, and is rather bigger...
#!/bin/ksh
# $ID$
# movefile
filename=$1
mv $filename.3 $filename.4
mv $filename.2 $filename.3
mv $filename.1 $filename.2
mv $filename $filename.1
--
"Intel engineering seem to have misheard Intel marketing strategy. The
phrase was ``Divide and conquer'' not ``Divide and cock up''"
-- <[EMAIL PROTECTED]> Alan Cox
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/lsf.html>
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: comp.os.linux.x
Subject: Re: GL/Mesa with X11
Reply-To: [EMAIL PROTECTED]
Date: Fri, 30 Jul 1999 00:06:10 GMT
On Thu, 29 Jul 1999 12:32:35 -0700, David
<[EMAIL PROTECTED]> wrote:
>I have recently upgraded my X11 package with 3.3.4 version of X. It
>says in the readme that the TNT2 chipset is now supported and I'm going
>to buy a Diamond Viper V770 card now. What I would like to know is
>exactly what the process would be to configure X to use GL/Mesa?
>
>In an earlier post there was a mention of going to the nvidia site and
>downloading their driver software for full support. Do I still have to
>do this now that X11 is supporting the chipset or does X11 provide the
>full capable support? Is there another procedure I need to do, such as
>just go straight to the mesa3d site and download the packages there?
>
>Mainly, I just want to get the full potential use out of my card but am
>looking for one good way to do it. If anyone has advice or suggestions
>please let me know. Thanks.
At this point, the probable "best way" to get the fullest "use" out of
your card is to avail yourself of one of the commercial OpenGL
implementations, either from Metro Link <http://www.metrolink.com/> or
from Xi Graphics. <http://www.xig.com/>
They can provide certified OpenGL implementations, and have tended to
be quicker to support new cards than XFree86.
XFree86 should be improving substantially in OpenGL support in version
4.0, but that's still real alpha at this point.
--
It is usually a good idea to put a capacitor of a few microfarads
across the output, as shown.
[EMAIL PROTECTED] <http://www.hex.net/~cbbrowne/xwindows.html>
------------------------------
** 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.misc) 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-Misc Digest
******************************
