This is the mail archive of the cygwin-developers@cygwin.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: fork and mutexs


On Wed, 2001-09-12 at 02:38, Jason Tishler wrote:
> Rob,

> My strategy is to change Cygwin (at least temporarily) to warn instead
> of abort.  And, then go to the python-dev list and attempt to fight that
> battle.  Do you concur?

A silent approach is better - otherwise I can just imagine the cygwin
list traffic.

"Foo works perfectly on Linux, but I get all these warnings here about
forking with waiting condition variables. Why is cygwin broken?"

If python forks() and immediately exec's then it's not buggy, spec
speaking. If python forks() and then does other stuff... well thats not
so good... particularly if it used pshared mutex's or cond variables.
Although in a certain sense pshared syncronisation is easier, because
the new thread can simply be not given ownership of the mutex's and it
just has to try and acquire them...

BTW: Greg, if you'd like to try that speed comparison between
win32pthreads, and cygwin pthreads again, I think you'll like the
results...


Rob 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]