Hi, > Gesendet: Donnerstag, 12. Februar 2015 um 21:59 Uhr > Von: "Branko Čibej" <br...@wandisco.com> > An: dev@subversion.apache.org > Betreff: Re: svn commit: r1659397 - in /subversion/trunk: TODO > build/ac-macros/swig.m4 > > On 12.02.2015 21:55, br...@apache.org wrote: > > Author: brane > > Date: Thu Feb 12 20:55:13 2015 > > New Revision: 1659397 > > > > URL: http://svn.apache.org/r1659397 > > Log: > > Do not attempt to build bindings with Swig 3.0+. Current versions > > break the Python bindings. > > Heads up for the above; should be in CHANGES, probably. Several people > have been trying to get our Python bindings to work with Swig 3.0.x, > without success. Even though the Perl and Ruby bindings work at this > point with 3.0.5 (at least, they compile and the tests pass), I think > it's safer to tell people to use Swig 2.x for the time being, until the > issues with 3.x are resolved.
You surely know that much of your direct code use is from distributions, not individuals who build from source. And in their development version there often simply isn't such a choice. I have SWIG 3.0.5 in my distribution development tree and it is only going forward. The swig project is working on the python regressions, and I will patch the svn or swig source to get it all to work. Would you not rather want to state that some items are not supported with SWIG 3.0+ than erroring out altogether? (Which is incorrect by itself because the python regression did not occur between 3.0.0 and 3.0.2) Andreas