Axel Simon <axel.si...@ens.fr> writes: > On Nov 8, 2009, at 19:41, Andy Stewart wrote: > >> Hi Axel, >> >> Now i try to write some multi-process program, so i need use NamedPipe >> to communication with those processes (ParentProcess and >> ChildProcess). >> And i need use multi-thread to do concurrent in process, example, >> some thread to >> paint UI, some thread to communication with other process through >> NamedPipe. >> >> But i saw you write below notes on gtk2hs homepage: >> >> ------------------------------> notes start >> <------------------------------ >> A note of caution: Using Gtk2Hs together with the -threaded ghc is >> currently not supported. The reason is that GHC’s garbage collector >> might free Xlib or Win32 objects using an OS thread different to >> the one >> that is used to execute all other Gtk+ functions. The result will be a >> program that occasionally fails with an Xlib error or a crash on >> Windows. We are working on fixing this. >> ------------------------------> notes end >> <------------------------------ >> >> So it's a big problem in linux? When you said *occasionally* fails. > > It affects Linux as well as Windows. > >> Is it easy to fix? > > It's not clear what the best fix is. The problem is that finalizers > of objects are called by the GC using a thread that might be > different from the thread that runs the Gtk+ main loop. Indeed, the > Gtk+ main loop might be running while the GC finalizes some objects > that may thereby release Xlib or Win32 resources. This creates calls > to Xlib or Win32 that are concurrent with respect to whatever the Gtk > + main loop does. We need to avoid this and it is not clear what the > best way is. > >> Can you give me some suggestions before i write multi-threads program? > > The best way for now is to try to write a program that does not need > the -threaded flag of ghc, i.e. to write a program that does not use > blocking C calls. I don't know if that is an option when you need > named pipes. In any way, I think we can fix the problem before the > next release. If we don't use -threaded flag of ghc, we just can use `forkIO` (Haskell threads) `timeoutAddFull`, even `timeoutAddFull` isn't natural way to implement multi-threads.
If we don't calling Gtk+ C functions in forkOS, forkOS will works? If that's right, we can just forkIO with Gtk+ calling and forkOS with others non-Gtk+ calling. BTW, which way is natural way to fix this problem? GHC concurrent GC or gtk2hs side? If just wrap some gtk2hs functions will fix this problem, i want to stop current work and help you fix and test, i need refactory my project with multi-thread support. Cheers, -- Andy ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Gtk2hs-devel mailing list Gtk2hs-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/gtk2hs-devel
