Re: Locking on MINGW

[Stephen Brandon]

On Tuesday 26 February 2002 14:54, Richard Frith-Macdonald wrote:
> Anyone know why objc_mutex_lock/objc_mutex_unlock takes more than five
> times longer on windoze-xp
> than a linux 2.4 series kernel?

Take a look at the backend for Win32 in libobjc. I think some of the 
primitive locking functions used on win32 will be at fault. (ie slow). Sorry 
I am not an expert on these.

I did however write a version of NSConditionLock for Win32 (NT and upwards) 
because the backend for this does not exist in libobjc for the Win32 target. 
It seems to work really well in the SndKit. It would be nice if it could make 
its way into libobjc of course... but I imagine it is about as slow as the 
locking you mention above, if not worse...

Let me know if you want to try it.

Cheers,
Stephen Brandon
stephen@brandonitconsulting.co.uk