gcl-devel
[Top][All Lists]
Advanced

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

Re: [Gcl-devel] Troubles bulding gcl 2.7.0t3


From: Vadim V. Zhytnikov
Subject: Re: [Gcl-devel] Troubles bulding gcl 2.7.0t3
Date: Tue, 09 Aug 2005 23:53:00 +0300
User-agent: Mozilla Thunderbird 1.0.6 (Windows/20050716)

Camm Maguire works:
Greetings!  Great to hear from you Vadim!  Thanks so much for your
work with GCL.

"Vadim V. Zhytnikov" <address@hidden> writes:


Hi!

I have some troubles in building and installing gcl 2.7.0t3.

First problem is MAXPAGE value.  I can build gcl only
with --enable-maxpage=254*1024.  With 512*1024 configure
hangs eating 100% CPU on maxpage test.  With 1024*1024
and 2048*1024 configure is OK but build crashes with
some memory-related error message.



Currently maxpage must be a power of 2 and <= 1Gb on 32bit Linux
unless one has a red hat enterprise kernel, then one can go up to 2gb,
but only with t4.  In any case, please try t4, you might have better
luck as some work was put in related to recent gcc compiler changes,
etc.

Well, MAXPAGE 256*1024, 512*1024 and 1024*1024 works well to me
with GCL 2.6.7pre.  With 2.7.0t4 256*1024 and 512*1024 are OK but
1024*1024 failed during build (can't allocate memory beyond MAXPAGE).

On our todo list is a more flexible and robust memory allocation
strategy which does not constrain maxpages at compile time.  More n
this later.


Very nice plan!



All other problems are related to make install stage.

1.  Makefile contains reference to clcs_macros.lisp
which was recently renamed to gcl_clcs_macros.lisp.
The patch which fixes this is in the attachment.


Thanks!  Committed to cvs head.


2.  make install doesn't install main gcl script
to (correct gcl script is created but make doesn't copy it into
/usr/bin/gcl).  I've honestly tried to understand why and which
part of makefile is responsible for this but quickly get lost.

3.  Looking at make install log (attached) in the
hope to understand problem 2 I found two other suspicious
places:

a) rm -f gcl.exe (marked by ******)
b) some strange erroneous gcl session (marked by #########)

Probably both a and b do not influence final install
result since similar things can be observed in gcl 2.6.7pre
install log but maybe this is worth cleaning?



We really need to overhaul this.  Others have made similar reports.
Its really a question of whether we hav the time to do it right, or
want to patch what we have.  automake is where we should be going.


Since I already spent some time meditating on GCL's makefile and
things it is supposed to do I'll try do some cosmetic patch for t4.


--
     Vadim V. Zhytnikov

      <address@hidden>
     <address@hidden>




reply via email to

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