Day 21, worked 9:30 to 9:30.
Started looking at the 64 bit, version of the software. Uncovered a host of bugs not because its 64 bit instead of 32 bit, but rather because on 64 bit Windows the software gets installed into "c:\program files (x86)" instead of "c:\program files" and somebody never considered the possibility of executing a command line longer than 255 characters. Oh yeah, and he decided to do an unbounded sprintf so going over 255 characters caused a stack overflow crashing my software. Amateur hour. At least Microsoft's got a good remote debugger that works on 64 bit platforms, which is more than I can say from my experience with Suse Linux.
Other than that, things are looking ok. I'm still sick and completely exhausted and wondering when on Earth I can stop working 12 hours a day and weekends without a single day off in fucking 21 days.
Isolated a bug in Perl core today which was the root of the mysterious crashes in my multithreaded interpreter. That was cool. Modules are notoriously buggy, but to find a bug in the Perl core itself is pretty rare. It's like when you find a bug in the C/C++ optimizer. People always point fingers at the optimizer and 99.99% of the time it's you doing something stupid and not the optimizer. Submitted a patch via perlbug (RT 41318).
Going to bed early tonight, hopefully some rest will help me overcome the cold I have.