Linux-Misc Digest #683, Volume #21                Sun, 5 Sep 99 17:13:10 EDT

Contents:
  Re: how can i start non-root programs at init? ("Bruce Merry (Entropy)")
  Re: networking slows down (Janet)
  Re: Is is possible to store debug information from a seg fault? (Chris Butler)
  Re: Optimal Linux RAID Support? Questions. (David Cooley)
  Problem making a cross-compiler (Sasa Ostrouska)
  Re: Amiga, QNX, Linux and Revolution (Philip Kaulfuss)
  Re: any RAMDISK programs out there? ("Bruce Merry (Entropy)")
  Re: C programming (William Burrow)
  Re: Amiga, QNX, Linux and Revolution (Guy Macon)
  Re: General Rant from a Linux Newbie (Lizard)
  Re: General Rant from a Linux Newbie (Albert Ulmer)
  Re: General Rant from a Linux Newbie (Lizard)
  Re: General Rant from a Linux Newbie (Lizard)
  Re: General Rant from a Linux Newbie (Lizard)
  Re: Best Linux Distro? / Best GUI? (Navindra Umanee)
  Support Delphi For Linux ("Ecmel ERCAN")
  Re: C vs C++ for Open Source projects (William Burrow)
  2.0.x -> 2.2.x (was Re: Help please - configuring sound in kernel 2.0.38) (Paul 
Kimoto)
  Re: Amiga, QNX, Linux and Revolution (John or Jenn)
  Re: Best Linux Distro? / Best GUI? (Peter T. Breuer)

----------------------------------------------------------------------------

From: "Bruce Merry (Entropy)" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: how can i start non-root programs at init?
Date: Sun, 05 Sep 1999 20:53:52 +0200

[EMAIL PROTECTED] wrote:
> 
> How do I set up the init-scripts so that some programs / daemons etc are
> run under a certain user's uid instead of roots uid? This has to be done
> before the corresponding user logs in.

Some daemons have a command line option, so look at the corresponding
help. For the rest, use su (look at the manpage for the appropriate
options). Since you are root in the init scripts you can use su without
the password.

Bruce 
/--------------------------------------------------------------------\
| Bruce Merry (Entropy)            | bmerry at iafrica dot com       |
| Proud user of Linux!             | http://www.cs.uct.ac.za/~bmerry |
|         /earth is 99% full. Please delete anyone you can.          |
\--------------------------------------------------------------------/

------------------------------

From: Janet <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.networking
Subject: Re: networking slows down
Date: 05 Sep 1999 12:15:02 -0700

[EMAIL PROTECTED] (Bob Tennent) writes:

>  >> I have been recently been experiencing a problem with my networking
>  >> slowing down.  If I ping my other machine, the ping time is normally less
>  >> than 1 ms.  However, occasionally (it has happened 2 or 3 times in the
>  >> last week), it becomes a lot slower, sometimes taking up to 30 ms.
>  >> However, if I just restart networking (using the network startup
>  >> script in 
>  >> /etc/rc.d/init.d), it goes back to the sub-1 time.  Any ideas?
>  >> 
> I've been having similar problems.  Which kernel and distributions
> are the people having these problems with?  I'm using 2.2.12 and
> RedHat 6.0, with many rawhide upgrades.

I'm using kernel 2.2.10 and RedHat 5.1, but I have glibc 2.1 and many 6.0
packages.

Janet

------------------------------

From: [EMAIL PROTECTED] (Chris Butler)
Subject: Re: Is is possible to store debug information from a seg fault?
Date: 5 Sep 1999 20:55:00 +0100

[comp.os.linux.misc - Fri, 3 Sep 1999 13:33:59 +0100] * Spike! wrote *
>> Do you have a core file ? Maybe you find a kind soul who'll help you debug
>> using it ...
>> (It might be at /core or somewhere like this).
> If all the filesystems are unmounted when the segfault occurs...
> Where would the core file go?

Where would the executable be read from?

-- 
Chris Butler
<[EMAIL PROTECTED]>

------------------------------

From: David Cooley <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux,comp.os.linux.hardware,comp.os.linux.setup
Subject: Re: Optimal Linux RAID Support? Questions.
Date: Sun, 05 Sep 1999 16:23:16 -0400



[EMAIL PROTECTED] wrote:

> 
> It seems the RAID code was pulled from the released 2.2.12 code,
> if I understand what was being discussed on the RAID mailing list.
> On the one hand some folks say "it's been stable for a long time"
> yet it appears Linus wasn't happy enough with the patches being in
> the kernel so they were pulled.  Not scheduled to re-appear until
> a future 2.3.x flavor.  Some one please jump in here if this isn't
> exactly right!

The patches available for the raid code still don't run with 2.2.12...
Been trying it.  Finally downgraded to 2.2.5 and it's running fine.
I have 2 meta disk groups of 24G each under RAID5.  they give me 20G
useable.

My hardware config is:
SUN UltraEnterprise 2
dual 300MHz UltraII CPU's
256 Meg ram
dual internal 9G drives (System)
external SUN D1000 StorEdge array with 12-4.2G drives.

I'm currently running the raid code and SAMBA.  It has become the home
file server!

------------------------------

From: Sasa Ostrouska <[EMAIL PROTECTED]>
Subject: Problem making a cross-compiler
Date: Sun, 05 Sep 1999 21:59:12 +0200
Reply-To: [EMAIL PROTECTED]

Hello !

                Can anybody help me to build a cross-compiler ? I'm
trying and i get the following error but don't understand why ?


make[1]: Leaving directory `/usr/src/buildegcs/texinfo'
Building the C and C++ compiler
make[1]: Entering directory `/usr/src/buildegcs/gcc'
rm -f tmplibgcc2.a
for name in _muldi3 _divdi3 _moddi3 _udivdi3 _umoddi3 _negdi2 _lshrdi3
_ashldi3 _ashrdi3 _ffsdi2 _udiv_w_sdiv _udivmoddi4 _cmpdi2 _ucmpdi2
_floatdidf _floatdisf _fixunsdfsi _fixunssfsi _fixunsdfdi _fixdfdi
_fixunssfdi _fixsfdi _fixxfdi _fixunsxfdi _floatdixf _fixunsxfsi
_fixtfdi _fixunstfdi _floatditf __gcc_bcmp _varargs __dummy _eprintf _bb
_shtab _clear_cache _trampoline __main _exit _ctors _pure; \
do \
  echo ${name}; \
  /usr/src/buildegcs/gcc/xgcc -B/usr/src/buildegcs/gcc/ -O2
-DCROSS_COMPILE -DIN_GCC    -g -O2 -I./include  -fPIC -g1  -DIN_LIBGCC2
-D__GCC_FLOAT_NOT_NEEDED   -I. -I/usr/src/egcs-1.1.2/gcc
-I/usr/src/egcs-1.1.2/gcc/config -c -DL${name} \
      /usr/src/egcs-1.1.2/gcc/libgcc2.c -o ${name}.o; \
  if [ $? -eq 0 ] ; then true; else exit 1; fi; \
  i586-pc-linux-gnu-ar rc tmplibgcc2.a ${name}.o; \
  rm -f ${name}.o; \
done
_muldi3
/usr/src/egcs-1.1.2/gcc/libgcc2.c:41: stdlib.h: No such file or
directory
/usr/src/egcs-1.1.2/gcc/libgcc2.c:42: unistd.h: No such file or
directory
make[1]: *** [libgcc2.a] Error 1
make[1]: Leaving directory `/usr/src/buildegcs/gcc'
make: *** [cross] Error 2

I'm using the slackware 4.0 with egcs-1.1.2 . Pls. if somebody can help
or if you need additional informations contact me.

Thank you in advance.
Sasa


------------------------------

From: [EMAIL PROTECTED] (Philip Kaulfuss)
Crossposted-To: comp.os.qnx,comp.sys.amiga.misc
Subject: Re: Amiga, QNX, Linux and Revolution
Date: Sun, 5 Sep 99 20:56:33 GMT
Reply-To: [EMAIL PROTECTED]

A potato resembling Paul E. Bell miraculously spoke thus:

> From this, I would say it sounds (in English) like "Leen'ooks" (ee as in
> clean, oo as in book, with the accent on the first sylable).  Thanks,
> now if I could get the guys at work to pronounce GIF as "jiff", rather
> than "giff"...

I would have thought the latter pronunciation was correct, seeing as the G
is for "graphics". Certainly, everyone I know pronounces it that way.

--
                 ,-- Email: change nospam to boehme ------.
 Philip Kaulfuss | Website: http://www.boehme.demon.co.uk |
                 `---- ICQ: 21755556 ---------------------'


------------------------------

From: "Bruce Merry (Entropy)" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: any RAMDISK programs out there?
Date: Sun, 05 Sep 1999 21:03:41 +0200

> Does someone know of any programs that set up a ramdisk (i.e. portion of
> the memory that is accessed like a filesystem)? Preferably, the program
> should take care of copying some files of my choice to the ramdisk when
> it starts and move them back to hard disk before the computer shuts
> down. (Of course, this could also be done with a separate script.)
> Ideally, it would also have compression (such ramdisks existed in the
> days of computers that ran from floppy), but that is not necessary, I
> have plenty of spare RAM.
> 
> System is RH5.2, KDE, kernel 2.2.

Linux has very good builtin ramdisk support (you need to compile it into
your kernel; I assume you can recompile your kernel if you've upgraded
it to 2.2). You could probably configure stuff in your scripts to create
and mount the ramdisk copy stuff into it, then copy stuff back out when
you shut down. You won't be able to access it compressed, though. If you
need this as a way of having a virtual disk (and the speed gain that
memory offers isn't the issue), then take a look at the loopback device
(in particular, see the man page for losetup). It is similar to a
ramdisk, but the image is stored on the hard drive instead of in memory.

Bruce
-- 
/--------------------------------------------------------------------\
| Bruce Merry (Entropy)            | bmerry at iafrica dot com       |
| Proud user of Linux!             | http://www.cs.uct.ac.za/~bmerry |
|         /earth is 99% full. Please delete anyone you can.          |
\--------------------------------------------------------------------/

------------------------------

From: [EMAIL PROTECTED] (William Burrow)
Subject: Re: C programming
Date: 5 Sep 1999 19:32:49 GMT
Reply-To: [EMAIL PROTECTED]

On Sun, 05 Sep 1999 16:46:59 +0200,
Utilisateur Red Hat Linux <[EMAIL PROTECTED]> wrote:
>I have read in the K&R something about conversion that i don't
>understand:
>if an int is coded on 16 bits and a long in 32 bits;
>-1L<-1U because 1U has become a signed long

Consider an unsigned char of 8 bits.  Then -1 would indicate the value
255, because negative numbers don't exist for unsigned numbers.  For an
unsigned 16-bit int, -1 would denote 65535.  See a pattern here (look at
powers of 2).  To compare the unsigned int to a long, the int must be
made the same length as the long.  What is being said, that this
conversion process makes the unsigned int into a signed long.  The
result is not equal because -1 is less than 255 -- the -1U was
deceptive, no?  

BTW, gcc doesn't seem to make these distinctions.


>and
>-1L>-1UL because -1L has become an unsigned long

This is a trick question, right?

>thnak you for helping.

Post to the comp.lang.c newsgroup for more appropriate information.
Better yet, go to the comp.lang.c FAQ on:  http://www.rtfm.org.


-- 
William Burrow  --  New Brunswick, Canada             o
Copyright 1999 William Burrow                     ~  /\
                                                ~  ()>()

------------------------------

From: [EMAIL PROTECTED] (Guy Macon)
Crossposted-To: comp.os.qnx,comp.sys.amiga.misc
Subject: Re: Amiga, QNX, Linux and Revolution
Date: 05 Sep 1999 12:50:20 PDT

In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] (John T 
Maguire) wrote:

>>It's Gif, not jif!!
>>
>>(Unless, of course, one chooses to pronounce "graphic" [the first word
>>in "gif"] as "jraphic", perhaps? :) ).
>>
>The creators of the gif format insist it's `jif'

So?  The creator of Linux (Lih-Nucks?  Lih-Nicks?) insists that it's
Lee-Nooks. The people who fought the Nazis didn't call them Not-Sees or
Not-Zees; they called them Nat-Sees (National Socialists).  The creators
of SCSI insist that it's "Sexy", not "Scuzzy".

Gig-uh-bite vs. Jig-uh-bite, anyone? ;)


------------------------------

From: [EMAIL PROTECTED] (Lizard)
Crossposted-To: comp.os.linux.advocacy
Subject: Re: General Rant from a Linux Newbie
Date: 5 Sep 1999 19:02:55 GMT

[EMAIL PROTECTED] (Matthias Warkus) wrote in
<[EMAIL PROTECTED]>: 

>It was the 5 Sep 1999 04:54:05 GMT...
>..and Lizard <[EMAIL PROTECTED]> wrote:
>> First task, of course, is to find a decent newsreader. It appears
>> there aren't any, at least if I want to use something a little more 
>> sophisiticated than trn, tin, slrn, or other 'cat walking on the
>> keyboard' inspired names. Those were lovely in 1980. This is almost
>> 2000. 
>
>Huh? How is lack or abundance of eye candy somehow indicative of how
>modern or not modern something is?
>
It's not. But here's what I can do with a Windows/Mac newsreader I can't 
(or don't know how, I shall be honest) do with tin/trn/slrn/etc:

a)Have multiple groups open concurrently.

b)Have multi-part binaries auto-assembled into single line items;queue 
several for DL+uudecode;have that process running in the background while I 
read/post.

c)Archive messages to folders.

d)Have multiple sigs, user ids, etc, on a group-by-group basis.

e)Have windows for different newsserves (I use 3 on a regular basis) open 
concurrently, with newsgroups from each displayed.

f)Cut&paste from my newsreader into my mailer trivially (or any other app, 
for that matter)

I am doing all this with XNews, a freeware newsreader that is barely a 
megabyte is size...hardly a 'bloated monster' in this day and age. There's 
no reason that Linux, with much better threading/multitasking, couldn't 
have a superior GUI newsreader.

>I'll give you a hint. Those newsreaders (which, except for trn maybe,
>did not exist in 1980 BTW) are very modern and usually provide better
>functionality than most of the eye-candy monsters. Has 
>
Here is my typical 'news day':
Open up XNews. Shift-F5 to update all groups. As soon as the first group is 
done, I open it and begin reading while the rest update in the background. 
At some point, I will stop reading this group (though with the window left 
open) to check out a binaries group. Oh goody, a hundred new pictures from 
some anime series. Click to queue them all (note that messages with missing 
parts are conveniently hidden), and then start decoding. They are all saved 
to my anime directory, of course. Meanwhile, back to reading. Hm. I need to 
repeat a point I made last week. Open my sent messages folder, cut&paste 
a paragraph into my reply, send. Hey, this looks like a good URL. Double-
click, launch Netscape. Interesting. This looks like it might be of 
interest to someone in rec.games.frp.misc. Copy it from netscape, open 
rgfm, post new message, paste, send. Etc.

Go on. Tell me I can do that in tin, with the same ease, speed, and LACK of 
having to remember a string of alphabet soup commands. Hell, tell me I can 
view two groups at once, or start a download operation in one group while 
I'm reading another.

I've been on Usenet since 1993 (not impressive by some standards, but I'm 
harldy a newbie). I've used tin+vi. It was nice. But it doesn't compare to 
a sophisticated GUI newsreader.

There's no reason this coudln't exist under Linux. The question is why it 
doesn't.

(I should note my very-short-term memory is not very good. Having to flip 
between screens constantly is not easy for me...having text displayed in 
mutliple concurrent windows is VERY helpful. This has nothing to do with 
'eye candy' -- text windows would be fine.)


------------------------------

From: Albert Ulmer <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.advocacy
Subject: Re: General Rant from a Linux Newbie
Date: Sun, 05 Sep 1999 21:41:26 GMT

> I just d/l and install the latest version of Gnome and
> so far it has been quite stable on my SuSE 6.1 system.
> Time will tell but first impressions are that it is
> light years ahead of the version that shipped with
> RH 6.0.

No surprise there, SuSE 6.x with any desktop environment has proven=20
far more stable on my systems than the RedHat 6.0 release.




------------------------------

From: [EMAIL PROTECTED] (Lizard)
Crossposted-To: comp.os.linux.advocacy
Subject: Re: General Rant from a Linux Newbie
Date: 5 Sep 1999 18:47:12 GMT

[EMAIL PROTECTED] (Bill Unruh) wrote in 
<7qubj9$7q7$[EMAIL PROTECTED]>:

>In <[EMAIL PROTECTED]> Steve Gage <[EMAIL PROTECTED]> writes:
>
>>Netscape is quite probably the most fragile program I've run under
>>Linux. You probably didn't have to power down, though. Ctl-Alt-Fx to
>
>I just had the priviledge of using it under Solaris. It is just as
>fragile there. It is Netscape, not Linux.

Recommendations, other than lynx, then? (I wonder if there will ever be a 
Linux version of Opera...)

------------------------------

From: [EMAIL PROTECTED] (Lizard)
Crossposted-To: comp.os.linux.advocacy
Subject: Re: General Rant from a Linux Newbie
Date: 5 Sep 1999 19:08:33 GMT

[EMAIL PROTECTED] (teknite) wrote in <9e5uq7.sa2.ln@localhost>:

>One last thing, dump vim and use pico for your
>editor. It is much easier to use and if you need the power
>of vim you can always learn it later.

Except I've been using vi since 1993 and I *like* it. I even use it under 
Windows for some things. :) Once you're over the learning curve, it's very 
fast and powerful.

------------------------------

From: [EMAIL PROTECTED] (Lizard)
Crossposted-To: comp.os.linux.advocacy
Subject: Re: General Rant from a Linux Newbie
Date: 5 Sep 1999 19:09:38 GMT

[EMAIL PROTECTED] (Donovan Rebbechi) wrote in
<[EMAIL PROTECTED]>: 

>>More rants as the situation warrents. BTW, how good is CodeWarrior as
>>an IDE? I used it on the mac and loved it, and it's for sale cheap for
>>Linux at my local CompUSA...recommendations/condemnations welcome.
>
>What kind of programming are you interested in ? 
>
>C++ ? C ? Java ?  Perl/Tcl/python ?

C++ and Java, mostly. 

------------------------------

From: Navindra Umanee <[EMAIL PROTECTED]>
Subject: Re: Best Linux Distro? / Best GUI?
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.admin
Date: Sun, 05 Sep 1999 19:49:06 GMT

Peter T. Breuer <[EMAIL PROTECTED]> wrote:
> I have news for you .. us admin types like to compile our own snapshots
> of gnome/kde (and this compilation of corba/mico for kde 2 is at times
> dipping into 240MB virtual memory as I speak ...), rather than have to put

It's *much* more reasonable without --enable-final, I've found.
--enable-final just seems to dump all the C++ code in one single file
and then attempts to compile the whole shebang in one go.

> (yay, let's hear it for Beta3).

Meaning Beta3 was one of the better releases?

> Oh .. the joys of finally upgrading your 2.7.3 gcc to 2.7.2.3 just so you
> don't have to fix the assembler in the mico 2.2.* compile anymore.

I found I had to use egcc instead of gcc for the linking otherwise
exceptions don't work properly and konqueror fails to run (everything
else works, though).

As for Mandrake, they have done a whole lot of tweaks and
optimizations that just makes the whole Linux experience better for
new users.  Mandrake is what Red Hat should have been.

-N.
-- 
"These download files are in Microsoft Word 6.0 format.  After unzipping, 
these files can be viewed in any text editor, including all versions of 
Microsoft Word, WordPad, and Microsoft Word Viewer."  [Microsoft website]
           < http://www.cs.mcgill.ca/~navindra/editors/ >

------------------------------

From: "Ecmel ERCAN" <[EMAIL PROTECTED]>
Subject: Support Delphi For Linux
Date: Mon, 6 Sep 1999 00:17:13 +0400

Hi,

Is there any newsgroup or website which we can discuss the so called Delphi
for Linux?

If not, I had set-up a place in my forum are at
http://avrora.virtualave.net/cgi-bin/dcforum/dcboard.cgi?az=list&forum=linux

If you know, another place please inform.

Ecmel ERCAN




------------------------------

From: [EMAIL PROTECTED] (William Burrow)
Crossposted-To: gnu.misc.discuss
Subject: Re: C vs C++ for Open Source projects
Date: 5 Sep 1999 19:42:54 GMT
Reply-To: [EMAIL PROTECTED]

On 05 Sep 1999 10:57:16 +0200,
Niels M�ller <[EMAIL PROTECTED]> wrote:
>The next issue is the religious one. Most people agree that C is a
>resonable language for writing typical unix applications. Well,
>perhaps we don't _like_ C, but we use it and we have learnt to live
>with it. On the other hand, there are many, otherwise quite
>reasonable, people, who really dislike C++, and won't touch a C++
>program except possibly if they get a large pile of money from it.
>
>So you'll have fewer potential hackers if you go with C++, even
>ignoring the possibility that there are fewer people who know C++ than
>C (I don't know whether or not the difference is large or relevant).

I don't suppose all those C++ Windows hackers left over when the world
dumps Windows would fill the void.  Of course, Java or some other
language may replace C++ as the Windows language of choice before that
happens.  Ha ha.  Ahem.

BTW, KDE is perhaps and example of a successful project based on a C++
library.

-- 
William Burrow  --  New Brunswick, Canada             o
Copyright 1999 William Burrow                     ~  /\
Step away from the pulpit,                      ~  ()>()
and nobody gets hurt....

------------------------------

From: [EMAIL PROTECTED] (Paul Kimoto)
Subject: 2.0.x -> 2.2.x (was Re: Help please - configuring sound in kernel 2.0.38)
Date: 5 Sep 1999 16:43:47 -0500
Reply-To: [EMAIL PROTECTED]

In article <[EMAIL PROTECTED]>, Stephen Cornell wrote:
> I don't want to install a 2.2.* kernel - I'm sure this would break a lot
> of applications and generally cause a lot of unnecessary work.

According to http://www.kernelnotes.org/change22.html (based on the
Documentation/Changes file, "Red Hat 5.2 ships with most necessary
software."

-- 
Paul Kimoto             <[EMAIL PROTECTED]>

------------------------------

Crossposted-To: comp.os.qnx,comp.realtime
Subject: Re: Amiga, QNX, Linux and Revolution
From: John or Jenn <[EMAIL PROTECTED]>
Date: Sun, 05 Sep 1999 19:48:24 GMT

In article <7qugrr$[EMAIL PROTECTED]> Guy Macon,
[EMAIL PROTECTED] writes:
>If you were writing code that
>kills people when it's wrong, (and an unexpected 1 millisecond delay might
>be enough to kill)

Ha!  I used to write code that was intended to kill people when it was 
right!  Maybe the wrong people, if it were wrong.  ;-)

------------------------------

From: [EMAIL PROTECTED] (Peter T. Breuer)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.admin
Subject: Re: Best Linux Distro? / Best GUI?
Date: 5 Sep 1999 20:18:22 GMT
Reply-To: [EMAIL PROTECTED]

Navindra Umanee ([EMAIL PROTECTED]) wrote:
: Peter T. Breuer <[EMAIL PROTECTED]> wrote:
: > of gnome/kde (and this compilation of corba/mico for kde 2 is at times
: > dipping into 240MB virtual memory as I speak ...), rather than have to put

: It's *much* more reasonable without --enable-final, I've found.

Yes. Your diagnosis is correct ...

: --enable-final just seems to dump all the C++ code in one single file
: and then attempts to compile the whole shebang in one go.

Well, optimize it in one go. Yep.

: > (yay, let's hear it for Beta3).

: Meaning Beta3 was one of the better releases?

It was.

: > Oh .. the joys of finally upgrading your 2.7.3 gcc to 2.7.2.3 just so you
: > don't have to fix the assembler in the mico 2.2.* compile anymore.

: I found I had to use egcc instead of gcc for the linking otherwise
: exceptions don't work properly and konqueror fails to run (everything
: else works, though).

Depends what you are after .. I have disabled exception handling via
builtin c++ stuff .. not only so that I can compile with gcc 2.7.2., but because
I'm not sure I trust it (I trailered the compilation in gcc 2.8.1 first too).
I don't run konquerer, and probably can't spell it :-(.

: As for Mandrake, they have done a whole lot of tweaks and
: optimizations that just makes the whole Linux experience better for
: new users.  Mandrake is what Red Hat should have been.

I agree, though I sincerely doubt the efficacy of the -pentium optimization.

That can't help by more than about 3-5% at best, and on anything but a
pentium classic it's not a help at all, as the dynamic optimization on
686 machines between the dual pipelines will be better than compile time
tweaks.  I'd never use -pentium.

What is undeniable is that gcc 2.8.* produces code that can be twice as
fast in itself as gcc 2.7.2 -produced code.  And egcs 2.95 is
comparable, though buggier (depending on your point of view on whether
its the new compiler or the old source code with the bug ...). So it
doesn't surprise me that they can get a faster result if they try. The
important thing would be if they cut out a fair proportion of the
inane RH goofups. Simply _testing_ their stuff before sending it out
would have helped! 

I hope SuSE starts contracting out its services as a release testing 
agency. Those guys are thorough. But then practically anything would
be more thorough than redhat. RH are a commercial enterprise under
pressure, like MS, to produce features rather than reduce bugs. They
produce early and test late, using the users as betatesters. It is
entirely predictable and to be expected that RH releases will be
buggy. If it were not so, RH would be out of business. That's their
market. That's nothing against redhat. If you prefer features to
bugs, use redhat. I don't. I have to fix the bugs. It's work for me.


--
Peter

------------------------------


** 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
******************************

Reply via email to