On 3/4/02 9:34 PM, "Max Horn" <[EMAIL PROTECTED]> wrote: > [I planed to send this over one week ago, but many things got in my way, > sorry] > > > I think we should make a 0.4.0 distro release pretty soon. To prevent > problems like last time, I'd like to have the release process > formalized more than it is (which is not at all). I believe this will > benefit us all. I plan to add a section to the documenation/policy on > this eventually. Following some thoughts of mine, not yet complete. > > > I) Distro Release Frequency > =========================== > Full distro releases should occur roughly every 2 months. This > ensures the binary distro doesn't lag behind to much (see point III). >
Sounds good to me, the current binary distro is a bit lagging right now. > > II) Time table for distro release > ================================= > <snip> > > > > III) Binary Distro > ================== > For now, the binary distro will continue to be only updated on each > full distro release. In the future, we may try to come up with a > system that allows for greater update frequency. Any news on getting fakeroot working on Darwin/MacOS X? As I understand it, if we were able to use fakeroot, we could utilise the SF Compile Farms, correct? If we were, they could be used to build the packages for the binary distro, instead of laying it all upon you, Max (and others). Just to add to that, if we *could* get fakeroot working, an idea would be to automatically build packages that are added into CVS, however some things should probably come into effect before that would be viable: 1) Signing of packages by maintainers/commiters (GPG, PGP, etc). I'm not sure how that would occur, but since you need SSH access to commit now anyway, this might negate it. 2) Add another distro, ala Debian. I find the Debian system works quite well, although the releases are a bit slower than what you would want... However, the idea of having unstable as bleeding-edge would be good, and then packages that receive positive feedback are moved to 'testing', and then to stable. This would ensure that we don't have problems with the packages, like we do every now and then... > Some parameters: > 1) The build enviroment will be fixed (i.e. I'll mark the exact > machine/OS revision/compiler version). We will note these and other > parameter so that we can reproduce the build setting later on if > required. > > 2) For now, packages will probably still be built only by me. In the > future, we'll want to extend this to a couple of other people. > However, it's mandatory that these be team members, available, and > guranteed to be trusted. We can't just take binaries from everybody, > for security reaons. Identify of people submitting packages will have > to be verified before their binaries can be put into the bindist. > PGP/GPG signing, perhaps? > > I'd like to encourage others to do likewise. BTW; you don't need a > full clean system for this; you can also rename your /sw dir for the > testing period, and bootstrap a new one (you can restore you real sw > dir back later on this way). The idea of this phase is to catch any > stable-only problems (like dependencies that can only be fullfiled in > unstable, etc., happens easily). > I'd be happy to help out with this, ... I've got 12Gb or so left on a drive in my G4/400 (used to be a bit of a speed demon, but not any more... :) > > UPDATE: Finally, thanks to recent events, I am now always testing a > stable system on an external Dual-800 G4. More on this in another > mail soon. > Woohoo! That's great! > > Cheers, > > Max _______________________________________________ Fink-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/fink-devel
