[M3commit] CVS Update: cm3

Jay K jay.krell at cornell.edu
Fri Jul 24 14:27:29 CEST 2009


Two good ways to speed this up:
 
build and install libgmp libmpfr, non-shared (if you are going to distribute the cm3cg)
don't optimize it
 
It seems that building libgmp and libmpfr are a tremendous fraction of building cm3cg.
 
Or what you did and have in mind -- only rebuild it when it changes.
In fact, more generally, don't make a new CVS checkout for every test run, and don't clean, or at least never clean m3cc, there is purported full incremental logic in there.
Perhaps pick a day of the week per machine where the CVS checkout is deleted and recreated.
 
 - Jay


----------------------------------------
> Date: Fri, 24 Jul 2009 14:19:49 +0000
> To: m3commit at elegosoft.com
> From: wagner at elego.de
> Subject: [M3commit] CVS Update: cm3
>
> CVSROOT: /usr/cvs
> Changes by: wagner at birch. 09/07/24 14:19:48
>
> Modified files:
> cm3/m3-sys/m3cc/src/: m3makefile
>
> Log message:
> add support for omitting unnecessary m3cc builds during regression tests:
>
> If WORKSPACE is set in the environment and PREBUILT_CM3CG is set,
> this file is just used as the build result. No further checks are
> performed.
>
> This is of course a hack, but our current resources of build servers
> are heavily loaded by continually compiling the cm3cg backend, which
> is not necessary in most cases. We can detect via CVS when it _is_.
>


More information about the M3commit mailing list