…er) when creating flavour dir. Also use rsync when syncing from TUV (it's a *lot* faster).
a couple run of the mill bugfixes. Still using a text format flatfile database. Maybe something fancier in the future.
why these are f14 packages instead of f13 or f15 I know not. Presumably the long term mingw32 solution will be a minimal bootstrapping procedure as per the recent sl-devel discussion with link to fedora packager providing patches and instructions to centos. And using f13 to minimize the sprawl of bootstrapping dependencies (unless f14 really was needed for some obscure reason)
recent cleanups caused some locations to get confused, this will get your netinst and DVD installer .isos back in the final user-visible/usable outputdir again. Also several comment evolutions, and a key for my zyx-liveinstaller rebootless liveos installer package.
- bake: enhancement: use dynamic httpd port to allow concurrent bakes - stage: untested enhancement: use repochart, closer to usable - bake: cleanup: simplify tmpdir usage - v-bake: cleanup: refactor/clean pungi-and-deps rebuild and install - pre-bake-clean: new script: offloading and enhancing some bake prep steps - defaults: s/x_builder_/x_bake_/g for consistency - all over: general trivial cleanups and comment evolution
I had briefly thought this might live as a standalone tool alongside el-build, however given the forthcoming integration with repochart to handle the pointrelease and updates binning, it doesn't make sense to try to do that anymore.
seems to be working well enough, got grub update finally built on build.ascendos.org... yippee!
- everywhere: general comment/formatting evolution/settling - defaults: factored/fixed more settings, x_bake_bug_url, etc... - generate-glibc32-koji-helper: more/better/correct component list - update-net-cache: new setting: x_i_can_justify_this_network_bandwidth_to_myself - v-bake: more/correctly use x_bake_distover setting - v-bake: undo mock site-defaults.cfg until qa scripts catch their justifications - v-bake: fix generate-glibc32-koji-helper's invocation and its output repo usage
don't make the output directory, just its parent
This script automatically generates the 'fake' glibc rpm that seems to be needed by koji to build some packages, like grub. See- http://lists.fedoraproject.org/pipermail/buildsys/2011-January/003501.html http://lists.fedoraproject.org/pipermail/buildsys/2010-May/003085.html http://lists.fedoraproject.org/pipermail/buildsys/2010-May/003085.html http://koji.fedoraproject.org/koji/buildinfo?buildID=153452
this unifies two prior scripts, and gets invoked by bake immediately after update-net-cache is called.