Hi,
I've been experimenting with a combined repo
(https://github.com/koenkooi/Angstrom-integration-layer/) the past few days and
I have some mixed feelings about it. I tried to mimic the yocto setup and the
questions I get from people are "how do I know where all the stuff is
originally from", "what if I want to contribute upstream" and "git submodules
are awesome, combo repos aren't".
So let's assume that we want a single git repo that people can clone the way
they expect (awesome, submodules ruled out already), is intuitive to use and
facilitates contributing back upstream. Since this is a from scratch effort I
don't need to mimic yocto since, but can if it makes sense. Having said all
that, here's the layout I was thinking of:
sources/
o bitbake (external git, bitbake)
o layers/
* meta (external git, oe-core)
* meta-openembedded (external git, meta-oe)
* meta-efl (external git, meta-oe)
* meta-angstrom (external git, meta-angstrom)
* ...
o downloads (DL_DIR)
build/
o tmp(-$TCLIBC) (TMPDIR)
o shared-state-cache (SSTATE_DIR)
output/
o images (DEPLOY_DIR_IMAGE)
o packages (DEPLOY_DIR_(IPK/RPM/DEB)
conf/
o auto.conf bblayers.conf local.conf sanity_info site.conf (under
git control)
All the meta-* directories are at the same level and the build output is easier
to find and not hidden a few levels deep in build/TMPDIR. I think this would be
an improvement over the existing angstrom and yocto layout, but I've been doing
this for the past 7 years so I'd appreciate a fresh look from other people[1].
regards,
Koen
[1] especially if you have been or are presently using eclipse, I don't get
along with IDEs
_______________________________________________
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
