[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: 1.2.0-rc1 tarballs up for testing/signing

From: John Lenz <lenz_at_cs.wisc.edu>
Date: 2005-04-06 19:42:41 CEST

On 04/04/05 19:03:55, Ben Reser wrote:
> http://fornix.brain.org/subversion/
> -rw-rw-r-- 1 svnrm svnrm 6973827 Apr 4 19:46
> subversion-1.2.0-rc1.tar.bz2
> -rw-rw-r-- 1 svnrm svnrm 8520113 Apr 4 19:46 subversion-1.2.0-rc1.tar.gz
> -rw-rw-r-- 1 svnrm svnrm 11364041 Apr 4 19:50 subversion-1.2.0-rc1.zip
> md5sums:
> c96be2d7591b7ee260ac372fe1248dff subversion-1.2.0-rc1.tar.gz
> b99e49eea110ebfca338b62db5f5a2be subversion-1.2.0-rc1.tar.bz2
> d28d5f6fd84b9f4651e879af0b9500c6 subversion-1.2.0-rc1.zip
> sha1sums:
> 75138673c63b8e5015ff2f9ffdc6f27e49f4d65d subversion-1.2.0-rc1.tar.gz
> a64ae487bbb3e752e074c59e1f9665fc5e53bd7b subversion-1.2.0-rc1.tar.bz2
> e65c81a507a95ad5ad7645c776ab48b841602011 subversion-1.2.0-rc1.zip
> Please test and send me your signatures. Thank you.

I noticed that SWIG is still required to build the tarball. In my opinion,
removing the requirement that SWIG be installed to build the subversion
tarball should be something that makes it into 1.2. No code changes need
to take place to support this, it is only Makefile changes and build script
changes so it shouldn't be a problem of testing. I attempted to hack the
build system to support this, but the way build-outputs.mk is built with
those python scripts is so complicated I couldn't figure out how to do it.

In case the previous mail got lost (there were no replies), something like
the following needs to be added to the makefile:

SWIGLIBDIR=`$(SWIG) -swiglib`

        $(SWIG) -python -external-runtime path/swigpyrun.h || \
        ( cat $(SWIGLIBDIR)/swigrun.swg > path/swigpyrun.h && \
          cat $(SWIGLIBDIR)/python/pyrun.swg >> path/swigpyrun.h && \
          cat $(SWIGLIBDIR)/runtime.swg >> path/swigpyrun.h; )

and then have the swigutil_py.c file include swigpyrun.h instead of
including the swigrun.swg, pyrun.swg, and runtime.swg directly.

The main problem is dist.sh does not actually run configure or make, so I
don't know where to add this into the build system. You would like it to
be added so the rule gets added to build-outputs.mk so that during normal
development if an .i file changes you have the rules to rebuild the SWIG
files. But you also need to build these rules from dist.sh, so having a
separate Makefile with the SWIG rules would work better. What I'm thinking
here is gen_make.py dumps all the SWIG rules (that is, only the rules to
build .c files from .i files) into a build-swig-outputs.mk which
build-outputs.mk includes. Then dist.sh could run
make -f build-swig-outputs.mk SWIG=/path/to/swig
The only variable build-swig-outputs.mk should have is the SWIG variable,
and when building normally the toplevel Makefile.in would set it, and
otherwise dist.sh would look it up and set it.


To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Apr 6 22:25:52 2005

This is an archived mail posted to the Subversion Dev mailing list.