Wednesday, September 2, 2009

A cause of crashes found

After making some changes the Windows version started to crash for me, which led me to identify one reason why the program was crashing for others. Frankly, I don't understand why it wasn't crashing on Linux as well.

Anyway, I implemented a quick fix for this, which is now available in version The change is quite visible in the tag editor, where all the cells display "N/A" while saving the data. Not the nicest option but I hope it takes care of most crashes.

Those getting assertion failures are very likely to still get them. If those happen while scanning a file, it would really help to have that file sent to me, so I can fix the program.

Another important change is in the "trace" code, the one that gets activated when a crash is detected. This is much faster now than before, for the Windows users. (I really tested tracing only on Linux for the previous version, and it didn't occur to me that it might take so much longer to open a file on Windows than on Linux.)

There are several other fixes that should help in cases when the program doesn't start at all, as well as a few minor improvements.

The source code and the Windows binaries are already available as I type (note that you might need to get them from my ISP, as for some reason Sourceforge still shows the previous version), while the Linux binaries are getting built.

No comments:

Post a Comment