Re: [Axiom-developer] new problem compiling wh-sandbox

[Gabriel Dos Reis]

"Bill Page" <address@hidden> writes:

| On March 29, 2007 10:21 AM Gabriel Dos Reis wrote:
| > 
| > address@hidden writes:
| > 
| > | Martin,
| > | 
| > | In fact you've encountered the same problem I have been
| > | having trying to merge gold, build-improvements, and
| > | wh-sandbox.
| > 
| > I would be interested in understanding why I'm not seeing
| > the build failure.  This is important because it would be
| > good to have fixes from wh-sandbox in build-improvements.
| > 
| > Do you have initial guesses?
| > 
| 
| I have seen problems similar to what Martin describes when
| I build Axiom Gold (patch-49 and patch-50), build-improvements
| and wh-sandbox on the axiom-developer.org server. The same
| source codes build properly for me on SuSE 10.2.

All my boxes have been upgraded in the last two weeks to SuSE 10.2...
I'm installing solaris 11 on 64bit intel-based machine I just
acquired, once I'm finished I'll try testing there.

| The axiom-developer.org server runs as a shared virtual host
| using a variant of the Red Hat linux kernel I am not exactly
| sure the of details, but the upshot is that the linux memory
| management in this configuration is a little different. As a
| result it does not support the configure option
| 
|  --maxpages=256*1024
| 
| I have to manually build gcl with a smaller memory size. With
| gcl-2.6.8pre it is possible to use values that are not powers
| of 2 and it seems that almost any size less than 256*1024 will
| work. Right now I am using
| 
|   --maxpages=196*1024
| 
| and the build completes normally.

OK.

| I am not sure whether this is related to Martin and Tim's
| problem or not but it is my initial guess.

Anyway, it is useful configuration information to have.
Did we have Axiom buildable with another Lisp system we would have a
different angle from which to attack the issue...

-- Gaby