Arfrever Frehtes Taifersar Arahesis wrote on Fri, Apr 01, 2011 at 18:33:40 +0200: > 2010-12-31 12:01:57 [email protected] napisaĆ(a): > > Author: danielsh > > Date: Fri Dec 31 11:01:56 2010 > > New Revision: 1054087 > > > > URL: http://svn.apache.org/viewvc?rev=1054087&view=rev > > Log: > > Don't error on 'none: bad command or file name' when SWIG was disabled > > at configure-time. > > > > (Yes, I'm on Debian, thanks for asking.) > > > > * Makefile.in > > (check-SWIG): New target. > > (swig-py, swig-pl, swig-rb): Use new target to fail early. > > > > * build/generator/gen_make.py > > (Generator.write): Use the new target in build-outputs.mk, too. > > > > * build/ac-macros/swig.m4 > > (SVN_FIND_SWIG): Don't check for 'none', just set the thing to 'none'. > > As a side effect, the message to stdout is obliterated. > > > > Modified: > > subversion/trunk/Makefile.in > > subversion/trunk/build/ac-macros/swig.m4 > > subversion/trunk/build/generator/gen_make.py > > > > Modified: subversion/trunk/Makefile.in > > URL: > > http://svn.apache.org/viewvc/subversion/trunk/Makefile.in?rev=1054087&r1=1054086&r2=1054087&view=diff > > ============================================================================== > > --- subversion/trunk/Makefile.in (original) > > +++ subversion/trunk/Makefile.in Fri Dec 31 11:01:56 2010 > > @@ -702,6 +702,12 @@ install-man: > > install-swig-py: install-swig-py-lib > > install-swig-rb: install-swig-rb-lib > > > > +check-SWIG: > > + if [ "$(SWIG)" = "none" ]; then \ > > + echo "SWIG disabled at configure time" >&2; \ > > + exit 1; \ > > + fi > > + > > Release tarballs contain files pregenerated by SWIG. This revision probably > breaks support > for building SWIG-based bindings from release tarballs when SWIG is not > available. >
Thanks for pointing that out. The problem I attempted to fix was that, in my trunk working copies (which I, most of the time, configure --without-swig), running 'make swig-foo' would attempt to run a command called "none". (This is because configure sets SWIG, which normally points to a 'swig' binary, to 'none'.) I'm sure there is a cleaner solution somewhere, but I don't have any ideas right now. (I could look later for a better place to move this check to, but a quick browse through the Makefile hasn't unearthed it for me.) Do you have any suggestions as to a fix?

