I have things set up in the following way... Say you have a project foo, that is developed in house, then I would have packages/foo/foo.inc packages/foo/foo_local.bb packages/foo/foo_1.2.0.0.bb
the foo.inc holds the core of the build (obviously). foo_1.2.0.0.bb is an official release, and points to some server some where that hosts the file foo_1.2.0.0.tar.gz, and foo_local.inc (or equivalently, a foo_git, foo_svn, etc) is assumed to be a working copy. the recipe for foo_local has the variable DEFAULT_PREFERENCE = "0" to prevent bitbake from trying to use it. If you are doing active development, then you set PREFERRED_VERSION_foo = "local" in your local.conf file, to let it use the local version. Since I'm on a number of collaborative projects, I don't want OE/bitbake to fetch tip revisions of the code, and so foo_local build requires you to set FOO_SRC = /path/to/source. It can then either build out of that directory, or, if you support out of source builds, just know to point to that location for building. Phil On Tue, 2009-04-07 at 17:39 -0400, Matthieu Poullet wrote: > Hi, > > I'm working on an Intranet and have some internal SVN/GIT servers I > fetch from. > > AFAIUI all fetchers call "try_mirror" first resulting in wget trying > to fetch a tarball version of the repo. > If it fails, it then tries to do a real checkout from the repo > mentionned in the URI. > > This wget call uses SRC_TARBALL_STASH so that it's possible to > redirect it to an internal web server. > > I've set up a web server and set SRC_TARBALL_STASH but I don't want to > use this for the internal servers. > It's fine for stuff I don't want to manage myself, but otherwise it > means that I must: > 1 - commit changes > 2 - generate a new tarball > 3 - put this tarball on my web server > 4 - rebuild the recipe > > that's a terrible workflow. > > If I don't use tarballs at all, everytime I rebuild it, i must wait a > very long time until wget timeout and then fetches from the repo, > that's terrible too. > > What I'd like to have could be a list of IP addresses and/or domain > names which the fetchers check first and then bypass the try_mirror > step. > > But maybe I'm missing something or there's another way to do this, but > I haven't found it yet. > I've asked many times in #oe without luck, I hope maybe someone here > could help. > > Regards, > Matthieu. > _______________________________________________ > Bitbake-dev mailing list > [email protected] > https://lists.berlios.de/mailman/listinfo/bitbake-dev > > _______________________________________________ Bitbake-dev mailing list [email protected] https://lists.berlios.de/mailman/listinfo/bitbake-dev
