I has come to my attention that - since the update to vala from the 0.34.x series has made it into the fedora 25(+) repositories - every package that generates .vapi files during build might have done so incorrectly, because of the following bug, which is present in vala releases up to 0.34.3 (also including 0.35.1), but has been fixed in git: Functions that have a return type of a fixed-size array were not treated correctly - they instead are wrongly indicated to return the base type of the respective array (the array qualifier "[]" is missing). I reported this bug at https://bugzilla.redhat.com/show_bug.cgi?id=1398738 a week ago, but it has not received any attention so far. I have notified Rico Tzschichholz directly and he has fixed this bug in the meantime (0.34 branch: https://git.gnome.org/browse/vala/commit/?h=0.34&id=7e76f75e42e7ac45c896e81e97955c8b2089d7e9 and in master), but it is not yet part of a stable vala release. As far as I can tell, every affected package that has been rebuilt since the faulty vala releases have made it into the f25 / f25 repos might contain errors because of this bug, and every affected package that is rebuilt now might introduce a new faulty package into fedora repositories. I don't know how frequent fixed-size-array return types are in functions overall, but at least one package in the stable fedora repositories has been affected by this bug: NetworkManager(-glib-devel), where libnm-util.vapi contains (at least one) such error - which results in build failures for packages using the affected APIs. Once a fix for the bug is available in fedora (I guess a patch to the latest vala release containing the fix linked above, since - so far - no new release has been tagged upstream), packages that generate .vapi files will have to be rebuilt to pick up the fix and to correct any errors that crept in while a faulty vala version was in the repositories. I find it somehow curious that nobody else has picked this error up before, but it might still take some time to affect more package builds since this is a one-removed problem ... Fabio _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
