Re: About cleaning up the thread stuff

[Marius Vollmer]

Kevin Ryde <address@hidden> writes:

> Marius Vollmer <address@hidden> writes:
>>
>>    dead-locks might be detected
>
> I wouldn't worry about that.  I think the combinations of things a
> thread can block on will make it too hard to be sure a deadlock is
> occurring.  There might even be some weird apparent deadlock that's
> meant to be broken by a signal handler cancelling one thread or
> something.

Yes, true.  Maybe the debugger could still try to figure out whether
there is a dead-lock, or the program could explicitely ask for it.
Etc.

>>    mutexes can be recursive
>
> Yep, I think that's friendliest.

I think there is use for both kinds of mutexes.  Sometimes you want a
recursive one, sometimes locking twice from the same thread is an
error.

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405