On Wed, 22 Dec 2004, Bob Jamison wrote:
Bryce Harrington wrote:
Ishmal, perhaps we should make the new dependencies default to "off" for one release, before making them required in subsequent releases? This would make them available to those who wish to test them, yet not impose them on people who just want to do compile the code.
Bryce
Ok, but please make this a short release cycle. ;-)
Yeah, I was talking with Mental about defaulting them to off. I expected some pains with the build for a few days, and I guess that this is enough. I will change the default in configure.ac to OFF for now.
Cool, thanks. And good suggestion to make this a short release cycle. We may have to shuffle some tasks on the roadmap a bit to do that, but I think it would be worthwhile, we waited a bit too long last time. Perhaps we could shoot for something mid-January? Or would that be _too_ early?
The only caveat is that the focus of this release is gtkmm conversion, and while Jizzbug and I have been pounding on it pretty hard for the past week or so, we could really use some additional hands. Doing both the gtkmm and the scripting (and all the other tasks underway) may stretch us pretty thin, so coordinating all of that for a quick release may be a big challenge. We may want to do some prioritizing.
HOWEVER, I humbly ask/beg/beseech/plead that everyone use --with-python and --with-perl, and if they have problems compiling, at least try to get it to work. If nobody ever fixes the build for platform 'X', it never will work for that one. This goes especially for the testers. This definitely needs testing. This stuff is going to take a huge amount of work, and any help will be appreciated.
*Nod* Agreed. And even if we don't get sufficient testing for the first release, at least we'll have gotten some, so that when we switch it to default "on" in the following release, we'll have had at least a little testing under our belts.
Bob, can you create a list of some tasks that folks could help with?
Bryce