Hello, Thanks for specifying this G_IO_NVAL condition.
If I remember correctly, there can be passed an Out-Of-Bound character(/message) on a pipe. It's some kind of priority message, generated for internal pipe management purposes. I *think* that G_IO_PRI condition will be the condition that's fulfilled when receiving such an OOB character (but this is only based on my own -wild- assumption, not on any documentation I found on the subject). Kind regards, PhB Daniel Yek wrote: > At 11:07 PM 12/5/2006, Philippe Bertin wrote: > >> Hello all, >> >> This is a new thread, based upon a very recent thread in this list, >> and a some things I asked myself in the (recent) past when I >> programmed a g_io_channel watcher callback. >> >> I used, for my GIOChannel, the condition G_IO_IN, in order to have my >> callback called whenever there was any input data available. Although >> I could imagine when one would possibly get any of both the G_IO_PRI >> (I guess the socket's O-O-B condition is triggering it ?) or G_IO_HUP >> conditions, still I had wondered when one would possibly get >> G_IO_NVAL ? Maybe that last one is only applicable in case of >> nonblocking operations ? > > > Closing write-end of the pipe causes G_IO_HUP. > > Shutting down GIOChannel "without removing g_io_add_watch()" or > closing the read-end of the pipe generate G_IO_NVAL. > > I'm not quite sure G_IO_PRI. > > I found out by writing a test application to (more) systematically > test GIOChannel. > > >> I must've missed some documentation ? >> >> Kind regards, >> >> PhB >> >> ... >> >>> That web site is really useful, but it didn't point out that you >>> need to handle G_IO_HUP, G_IO_NVAL, (and maybe G_IO_PRI). Handling >>> these helped my troubles, but I don't know if these have anything to >>> do with your trouble. >>> >> >> -- >> Daniel Yek > > > > _______________________________________________ gtk-app-devel-list mailing list gtk-app-devel-list@gnome.org http://mail.gnome.org/mailman/listinfo/gtk-app-devel-list
