http://bugs.winehq.org/show_bug.cgi?id=14851
--- Comment #8 from Austin English austinenglish@gmail.com 2009-03-31 00:40:02 --- (In reply to comment #7)
Results of regression test between 0.9.38 and 0.9.34
2a9dfe242b8d2b26c3498f27002c7190f1d7e77f is first bad commit commit 2a9dfe242b8d2b26c3498f27002c7190f1d7e77f Author: Alexandre Julliard julliard@winehq.org Date: Thu Apr 5 11:29:30 2007 +0200
ws2_32: Don't access the IO_STATUS_BLOCK until the I/O operation is
finished.
:040000 040000 d860d19991e8bab957cfae519873abe38a057558 fecf8de913eb48ffdd3eb4a689f6ddf217fedcf7 M dlls ========
Try reverting the patch in git. Or build just before it: $ git checkout 2a9dfe242b8d2b26c3498f27002c7190f1d7e77f ; patch -p1 -R 2a9dfe242b8d2b26c3498f27002c7190f1d7e77f $ ./configure && make depend && make
This regession ended in a between a build that would open and connect, and a build that would crash on load. This ultimately does not answer the question as to what is causing the connection problem, but I am unable to dissect the two issues from each other.
You can, it's just long and hard. You have to test for one regression at a time.
http://wiki.winehq.org/ReverseRegressionTesting may help.