|
From: Vadim Z. <vz...@ze...> - 2011-12-12 12:41:58
|
On Mon, 12 Dec 2011 07:16:56 +0000 William S Fulton <ws...@fu...> wrote: WSF> > Unfortunately I still can't run the test suite for Java because of WSF> > the problem with doing it from a build directory that is different from WSF> > source directory, see this (very old) thread: WSF> > WSF> > http://thread.gmane.org/gmane.comp.programming.swig.devel/20043/ WSF> > WSF> Did you miss the last reply in that email chain (28 July 2010)? No, but it still doesn't work for me, in spite of what this email said (top_srcdir is still wrong, it tries to look in "../$HOME/src/swig" instead of just "$HOME/src/swig"). I never replied to that thread because I wanted to debug it to understand why did it work for you but not for me but finally never had time to do it. I'll try to look at it again... WSF> > But I did test minherit2, contract and director_frob manually and they all WSF> > generate correct (i.e. the same as before) code now. WSF> > WSF> > Sorry again for the breakage and thanks a lot for noticing it! WSF> WSF> Thanks for quick fix. Looks good now. Shame you don't have a small test WSF> case for the test-suite, but I understand it was hard to find such a WSF> test case. Yes, even now I don't really know how to produce one that would fail with the old code reliably as it depends not only on the order of the elements in the hash table initially but also on where the new elements are inserted in it. This is why this bug was so nasty, too, as it appeared and went away seemingly randomly in our code as completely unrelated things were modified. Regards, VZ |