Sunday, December 17, 2006
I have got two application level crashes, one severe, that are holding up my project.
And you know what? I'm Happy. Why? They are repeatable. Repeatable problems can be solved.
A couple weeks back I was plagued by a bug that seemed to occur randomly, under extreme memory pressure, and I finally gave up on finding the source and moved to a faster machine. That move has had good side effects, I now have two X86_64 boxes, one for production music, the other that I can build and run debuggable code on.
I still need to get an x86 box running, because I think the severe crash is probably a 64 bit issue, and only building an x86 box will disprove that theory.
But first I filed bug reports on rosegarden and ardour2. The code for these programs is fiendishly complex and better minds than mine are already working on them.
Still, more brain cells can't hurt.
I just finished rebuilding ardour2 for debugging, am now building rosegarden svn for debugging, and am installing the libraries required to finally get the x86 box (called monk) working too... which gives me another out for production. I'm really tired of x86_64 issues wherever I go, and this particular box has enough ram for the job, barely. (I might be able to cannibalize my other x86 box to get me up to 1GB)
And maybe I'll get a chance to fool with a few VST plugins.
And I think I'll re-re compile ardour2, logging it's output, for all I know the compiler's been issuing a warning on the problem since day one.
With three boxes doing builds madly, I guess I'll go work on some christmas cards.
UPDATE: Ardour2 on the x86 doesn't crash when I hit record on rosegarden (as it does on x86_64). It looks like I was right - it IS a 64 bit issue stopping me dead. Now that I have an x86 box running I can move off of x86_64 and continue on the project - or hack on the two bugs - which of those paths I take depends on which side of my brain wakes up first monday morning. ¶ 7:26 PM
And you know what? I'm Happy. Why? They are repeatable. Repeatable problems can be solved.
A couple weeks back I was plagued by a bug that seemed to occur randomly, under extreme memory pressure, and I finally gave up on finding the source and moved to a faster machine. That move has had good side effects, I now have two X86_64 boxes, one for production music, the other that I can build and run debuggable code on.
I still need to get an x86 box running, because I think the severe crash is probably a 64 bit issue, and only building an x86 box will disprove that theory.
But first I filed bug reports on rosegarden and ardour2. The code for these programs is fiendishly complex and better minds than mine are already working on them.
Still, more brain cells can't hurt.
I just finished rebuilding ardour2 for debugging, am now building rosegarden svn for debugging, and am installing the libraries required to finally get the x86 box (called monk) working too... which gives me another out for production. I'm really tired of x86_64 issues wherever I go, and this particular box has enough ram for the job, barely. (I might be able to cannibalize my other x86 box to get me up to 1GB)
And maybe I'll get a chance to fool with a few VST plugins.
And I think I'll re-re compile ardour2, logging it's output, for all I know the compiler's been issuing a warning on the problem since day one.
With three boxes doing builds madly, I guess I'll go work on some christmas cards.
UPDATE: Ardour2 on the x86 doesn't crash when I hit record on rosegarden (as it does on x86_64). It looks like I was right - it IS a 64 bit issue stopping me dead. Now that I have an x86 box running I can move off of x86_64 and continue on the project - or hack on the two bugs - which of those paths I take depends on which side of my brain wakes up first monday morning. ¶ 7:26 PM
