On Thu, Mar 26, 2020 at 06:45:46PM +0100, Stefan Sperling wrote: > On Thu, Mar 26, 2020 at 04:30:26PM +0100, Stefan Sperling wrote: > > On Thu, Mar 26, 2020 at 02:23:05PM +0000, Michael Pilato wrote: > > > Maybe it's just been too long for me, but I thought our releases and > > > release > > > candidates were supposed to have the SWIG-generated stuff included? I see > > > those files in, say, subversion-1.13.1.tar.bz2, but not in this new rc1 > > > drop. > > > > Yes those files should exist. > > > > Not sure what happened. I used release.py and didn't see anything in > > its output that indicated a problem. > > > > I will investigate. Thanks! > > Re-running release.py in verbose mode reveals: > > [[[ > /tmp/svn-rel/tempdir/subversion-1.14.0-rc1/subversion/bindings/swig/include/svn_containers.swg:57: > Error: Unknown directive '%hash_argout_typemap'. > *** Error 1 in /tmp/svn-rel/tempdir/subversion-1.14.0-rc1 > (autogen-standalone.mk:329 'subversion/bindings/swig/python/core.c') > Creating build-outputs.mk... > Wrote: subversion/libsvn_subr/errorcode.inc > Wrote: subversion/libsvn_subr/config_keys.inc > Creating svn_private_config.h.in... > Creating configure... > > You can run ./configure now. > ]]] > > So it looks like our autogen.sh doesn't do proper error checking :( > > I'll try to get that worked out and then roll rc2.
Whew, this took me a while, but I figured it out eventually. The py3 branch accidentally broke a temporary Makefile that is only used by autogen.sh. See http://svn.apache.org/r1875742 for a fix, which is now also nominated in STATUS. It is not surprising that we only noticed this now when we tried to roll a new release. Thanks a lot for catching and raising this problem, Mike! Given the nature of the problem I think we should get the fix merged to 1.14.x before a new release is attempted. Unless there are pressing concerns over timing, I will wait a few days for this and hopefully other patches to get merged to the 1.14.x branch before rolling rc2. Regards, Stefan