On Fri, Jun 20, 2003 at 09:20:49PM -0400, felix zaslavskiy wrote:
> Its hard figuring out what the problem with a threaded app could be.  

So you expeact Nathan to magically fix it - how?

Actually it's not very hard to figure out what the problem is. The assertions
fired by libpthread (causing a core in the default setting) will imediately
tell you what's gone wrong. You can't expect to see the messages if you run
mozilla detached from a terminal with stderr redirected to /dev/null, of
course.

It's sometimes a bit hard to figure out how to fix the application properly,
but you can't blame that on the OS or libpthread.

The only thread related bugs not in this league are the kernel panics. I only
see the SMP ones myself, there are reports of one not SMP related remrunqueue
panic, which is quite new. That's all. For me it runs rock solid on all
machines not having "options MULTIPROCESSOR" in their kernel config.

So, overall, to me, this looks not too bad. I had hoped that the (known) SMP
issues had been solved a lot faster. But then I had hoped to finish a lot of
my own NetBSD stuff much faster too.


Martin