On Tue, Sep 18, 2012 at 11:05:55AM +0100, Philip Martin wrote: > s...@apache.org writes: > > > Author: stsp > > Date: Mon Sep 17 16:29:25 2012 > > New Revision: 1386696 > > > > URL: http://svn.apache.org/viewvc?rev=1386696&view=rev > > Log: > > * configure.ac: Don't raise a fatal error if --with-gnome-kerying was passed > > but the necessary things can't be found. Just print a warning and move on > > without compiling GNOME Kerying support. > > Is that sensible behaviour? I think it's a bit odd to continue without > gnome keyring support if the user explicity requested it.
Before my change, there was no way of saying a) "please use GNOME keyring if you can find it, else, don't bother" It was either b) "don't compile GNOME keyring support" or c) "error out if it cannot be compiled in" With my change, we now have a) and b), but not c). I want to eventually compile GNOME keyring support on the bb-openbsd bot, because currently no bot is even trying to compile it, as far as I can tell. But I don't necessarily want the build on that bot to fail if the gnome keyring packages happen not to be installed (which they aren't yet). If you want c), we could look for GNOME keyring by default, enable it if found, and then change the semantics of --with-gnome-kerying back to "error out if gnome keyring isn't found". That would restore all possible behaviour options.
