[EMAIL PROTECTED] wrote: > New commits: > http://git.directfb.org/?p=core/DirectFB.git;a=commit;h=dcf2a2b7089fd02105f68bba5da3130c1d72f42c > commit dcf2a2b7089fd02105f68bba5da3130c1d72f42c > Author: Claudio Ciccani <[EMAIL PROTECTED]> > Date: Fri Jun 29 14:48:28 2007 +0200 > > Added a new Fusion/IPC implementation using standard system calls: > > - Skirmish gets implemented by using sched_yield()+usleep, exactly like > linuxthreads does with pthread mutexes, > but it's also capable of automatically unlocking whether the owner > process died without doing it. > > - Property works like Skirmish, but it doesn't sleep and it doesn't > check for a dead owner. > > - Dispatching and co. make use of raw unix sockets: there is a local > address specific to each fusionee > (/tmp/[EMAIL PROTECTED]@/@HEX_FUSION_ID@) and each call (/tmp/[EMAIL > PROTECTED]@/[EMAIL PROTECTED]@[EMAIL PROTECTED]@), > and messages are sent directly from the generator to the socket of the > listeners. > From a network point of view, this is similar to UDP Multicast (e.g. > conference or meeting). > > The new implementation gets build when --enable-bulti is specified but no > usable linux/fusion.h was found > (however you can edit fusion/build.h to force building it).
Wow :) How complete, usable and stable is it? -- Best regards, Denis Oliver Kropp .------------------------------------------. | DirectFB - Hardware accelerated graphics | | http://www.directfb.org/ | "------------------------------------------" _______________________________________________ directfb-dev mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
