[EMAIL PROTECTED] wrote: > > > Gerard Vermeulen wrote: > > > > > > Great Phil, > > > > > > but I really think that adding the Py_BEGIN_ALLOW_THREADS and > > > Py_END_ALLOW_THREADS to the wrappers may really help to run threads > > > more smoothly, in case that the wrapped functions ask a lot of CPU time > > > or require I/O. > > > > Currently this is done (via sipReleaseLock() and sipAcquireLock()) for > > calls that handle Qt events (eg. QApplication.exec_loop()). There > > shouldn't be any problem with adding calls to things like > > QFile.readBlock() etc. Or are you suggesting adding them for every > > single Qt call? > > > I did not realize that these macros were already handled via those > ..Lock() functions. I did just a few tests and to my surprise > multithreading worked better than I expected. > > Yes, I was thinking to adding them to every Qt call. Of course that > is a naive brute force approach. It should be better to add > a sip directive, allowing thread switching in the Qt calls that > may demand a lot of time.
The directive is /ReleaseLock/. I'll add it to all the read/write methods I can find. Phil _______________________________________________ PyKDE mailing list [EMAIL PROTECTED] http://mats.gmd.de/mailman/listinfo/pykde
