macrobug

My ultimate aim by capturing all this information is to produce a swish view of everything that’s happened to lead you up to your current breakpoint, across all threads and processes on the device. That’s really what I mean by “flow of control monitoring”. Everything I’ve done so far – the high-speed trace capturing system, the hooking into Eclipse APIs, the symbol resolution pain and even the Memory Tracer widget I came up with to test those subsystems – has been a prelude to getting this overall flow-of-control tracing working.

Unfortunately, the task is so complex that I’ve never really been able to convince myself whether or not the end product will actually work – or if there’s some logical or technical flaw which mean the results of my techniques will be useless.

Well, today, I finally got close enough to find out. I’ve pieced everything together enough to get the view to appear.

And indeed there was a logical flaw that means the whole thing is useless :-)

But I’m pretty sure it is solvable. More in the next couple of days…

This entry was posted on Thursday, February 1st, 2007 at 6:55 pm and is filed under Business. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.