[gaim-migrate @ 12431] " The following log snippets should explain it: " --rlaager (20:24:00) rlaager: Regarding the signal handling conversation the other day... I've written a patch to stop calling signal handlers and return as soon as we find one signal handler that returns TRUE to indicate that it's handled the signal. Is this the right approach? (20:24:22) Ethan Blanton (Paco-Paco): the trouble is that it's documented to behave exactly the way it does (20:24:31) Ethan Blanton (Paco-Paco): so changing it is notbackwards compatible (20:24:31) rlaager: I'm talking for HEAD. (20:24:41) Ethan Blanton (Paco-Paco): oh, I think that's a good approach, yes (20:24:53) rlaager: The way I've described is how I *expected* it to work, having not read the documentation. (20:25:09) Ethan Blanton (Paco-Paco): I'm convinced (20:27:04) Stu Tomlinson (nosnilmot): rlaager: this, I assume, breaks the generic-ness of signals, by assuming that any that return values return booleans? (20:27:26) Ethan Blanton (Paco-Paco): please break it (20:27:33) Ethan Blanton (Paco-Paco): we already have out-parameters (20:27:42) rlaager: nosnilmot: from what I can see, the return type is handled as a (void *)... so I'm checking that ret_value != NULL (20:27:57) rlaager: nosnilmot: that's the correct way to do it, right? ... (20:29:01) Ethan Blanton (Paco-Paco): allowing a meaningful return value is an over-engineering (20:30:07) rlaager: even after this patch, you should be able to return meaningful return values (20:30:15) rlaager: it'll just short-circuit on the first handler that does committer: Tailor Script <tailor@pidgin.im>