On 18 Apr 2010, at 15:34, Fred Kiefer wrote: > Even if we could all agree on the current patch, we should then give it > at least two weeks to be tested by a broader audience.
If we're considering delaying the release slightly, then please could people test both -gui and -back with the non-fragile ABI? I've been using it on FreeBSD without issues for a week or so, but Niels reported some problems with Linux. It would be nice if we could advertise it as a feature of this release. > The switch to cairo as the default backend was a step that I should have > done right after the last release. Doing so now is just not advisable. > Or is it? I thought we deprecated the xlib and libart a couple of years ago... Is there any reason not to make Cairo the default? It's the only one that I've been using for almost as long as I've been using GNUstep. On the subject of defaults, can we revisit the issue of exception handling? I noticed recently that we're still using the setjmp() / longjmp() model. Unlike Apple, we don't have code that makes this play nicely with the new exception handling syntax, so we can't mix code that uses @try and code that uses NS_DURING unless we compile with the zero-cost model. On 18 Apr 2010, at 17:34, Richard Frith-Macdonald wrote: > Why not just make *two* windows installer releases with the two different > user interfaces? Doing that ought to actually help advertise the theming. Sounds sensible (although it generates some more work for someone-who-isn't-me). Provide screenshots next to the two installers and let people download the one that they want. David -- Sent from my Apple II _______________________________________________ Gnustep-dev mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnustep-dev
