Ingo Weinhold a écrit : >On Thu, 28 Apr 2005, Jérôme Duval wrote: > > >>i just committed this change: >>https://lists.berlios.de/pipermail/haiku-commits/2005-April/000900.html >> >>There are instructions there for enabling input_server with app_server (with >>cursor comm). I tested with the windows test app (I even increased wait time, >>because app_server takes more than 1 sec to start, at least for me). >> >>I didn't find a way to not have absolute paths for input_server and view input >>device. Anyone, feel free to fix this :) >> >>As it seems Registrar isn't available when building with >>RUN_WITHOUT_APPSERVER=1, BMessageRunner is failing, this is why we can't >>use it atm >> >> > >Sure you don't mean RUN_WITHOUT_REGISTRAR? At least then I would >understand that BMessageRunner doesn't work. BTW, AFAIK there is no reason >not to run our registrar under R5 (you just have to build and run >obos_registrar). Last time I tested it, it worked, that is. > > In fact, it was REGISTRAR. I'll try to test with it then :) > > >>(BTW i had a problem with BEntry::GetParent() too). >> >> > >Care to enter the problem in bugzilla? > > I don't know if it comes from using RUN_WITHOUT_REGISTRAR atm. I think i saw a kernel call for getParent. I'll test it again. Bye, Jérôme