Release 0.5-RC3: Statistical bugfixing
The crash reports filtered in today, and at the end of the day I sorted them into folders. There were literally 26 identical crashes when playing VIDEO_TS, which is a great sign that a lot of people are running into the problem. Thanks so much to the people who wrote a quick description. Feel free to add your email if you don’t mind me contacting you for further information.
- FIX: Possible crash when adding or editing source.
- FIX: Crash playing VIDEO_TS files (the biggest one!)
- FIX: Code signing is done correctly, so it will behave with Leopard Firewall.
- FIX: Lots of crashes as a result of some optical drive code still being enabled by default.
- FIX: A *ton* of hangs doing various things (e.g. adding sources, stopping video). The problem was that the scanner thread (which sometimes starts when the program starts), creates a big-ass database transaction inside which it does tons of time-consuming things like network scraping. This means that (a) no other database operations can run, like saving where you were in a file for resuming when you stop, or setting content on a source and (b) if the scan gets aborted for some reason, you end up with nothing in the database! The transactional design is simply incorrect. For now I’ve simply disabled the transaction, which fixes all the hangs, but it *may* introduce other issues (but hopefully not). I’ll choose the devil I don’t know on this one.