From: Guenter B. <bar...@we...> - 2003-03-23 00:15:32
|
hallo there, > As far as I understand it, this can't be related to the change of > assert to XINE_ASSERT, since if XINE_ASSERT causes an abort, then > so would the regular assert. Also IIRC one of the two asserts has > been fixed. guess you are right here. there is still the option of changing XINE_ASSERT to check an environment variable or something so it would regularly only spit out a warning and stack trace and only abort() on request. once again i'd like to mention that gtk and probably also gnome went down this road and it seems to work quite well for them. they also have additional macros that test conditions and return if the condition fails, so something like g_return_if_fail (this != NULL); can become quite usefull. once again, these macros could optionally abort() which means that these macros would output valuable debug information to logs users send to the mailing list, while developers can make them abort() to investigate things in gdb. > Also, I reported breakage of the QuickTime code a while back, which > hasn't been fixed yet to the best of my knowledge, and this hasn't > stopped later releases from being made. Especially since the current > ffmpeg aborts seem to be with badly broken streams, I don't know that > I would hold up the release just for that (as far as I know the QuickTime > breakage I reported was with a valid QuickTime stream). my point exactly. there had been valuable bug fixes and enhancements in cvs and i think it is important to bring those to the users as quickly as possible - release early, release often. if the ffmpeg asserts get fixed or other important improvements are made, we can have beta10 in a few days. this also has the advantage of making the rapid progress the xine project is making (thanks to you, fellow developers! :) ) visible to end-users. keep up the good work, guenter |