pan-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Pan-users] Re: 0.115 compile notes and feature request


From: Per Hedeland
Subject: Re: [Pan-users] Re: 0.115 compile notes and feature request
Date: Tue, 3 Oct 2006 01:06:23 +0200 (CEST)

Douglas Bollinger <address@hidden> wrote:
>
>On Mon, 2 Oct 2006 09:47:54 +0000 (UTC)
>"Duncan" <address@hidden> wrote:
>
><snip>
>> However, there's apparently a bug in gcc 3.4.x in regard to something pan
>> does, and it's now an older compiler, with most distributions having
>> already moved to gcc 4.x (Gentoo, which I use, is now stable on 4.1.1 on
>> most archs including x86 and amd64), so it shouldn't be a huge issue.  As
>> well, a lot of distributions will have binaries available once 1.0 comes
>> out, if they don't already, so fewer will be compiling.
><snip>
>
>Well, I use Gentoo as well, Duncan, but I'm master of my box and still use the
>old gcc:
>
>address@hidden ~ $ gcc --version
>gcc (GCC) 3.4.6 (Gentoo 3.4.6-r1, ssp-3.4.5-1.0, pie-8.7.9)
>
>I just build Pan 0.115 straight from the tarball without any problems using
>the regular defaults and the resultant binary works perfectly fine.  
>
>Perhaps this is the problem while trying to build Pan on a 512M box:
>
>0292 doug      25   0  601m 591m 5060 R   99 29.2   1:01.46 0 cc1plus
>
>It's not a gcc or Pan bug, you just need some more RAM or grab binaries from
>someone that does.

Yes, but it's absurd that compilation, not of pan itself, but of a test
program (scorefile-test) that most users will never even run, should
require more than half a gig of RAM and several minutes of CPU time just
to get compiled with -O2 - especially when compiling it without any -O
flag reduces the RAM and CPU requirements for compilation by some 90%,
and the resulting program still completes in milliseconds if you
actually run it.

If I could find my way through the labyrinth of automake stuff I would
suggest a patch to remove the -O2 for this particular program, for now
I've settled for having a little script that tweaks the generated
Makefile between ./configure and make... - I've no problem at all
building the rest of pan (including test programs) with gcc 3.4.2 and
standard options on a box that occasionally can't provide the amount of
memory that this particular compilation needs.

--Per Hedeland




reply via email to

[Prev in Thread] Current Thread [Next in Thread]