future OpenEmbedded updates
gergely.kis at gmail.com
Sun Jun 10 08:45:40 CEST 2007
On 6/6/07, Koen Kooi <koen at dominion.kabel.utwente.nl> wrote:
> > I would like to propose is org.openmoko.dev branch in OpenEmbedded with
> > limited commit access to OpenMoko guys which will be able to add/change
> > stuff there and OE devs will not. This will also give possibility to
> > cherrypick bugfixes from org.openembedded.dev branch to OpenMoko one (and
> > vice versa) in clean way.
> That is, a branch *without* any overlay of any kind.
I like your idea, but I think it is a man power issue at the side of
the OpenMoko team. A "base platform" integrator would be needed whose
role is to track / sync with the main OpenEmbedded branches. The best
person for this job would be of course a core OpenEmbedded contributor
with a contract with FIC...(hint, hint)
Also I think it is unrealistic that OpenMoko only uses Monotone for
version control, and I support their decision to go with a local
repository as well. I think this can be seen as a kind of strategy: a
platform can't commit to a single version control solution, because
too many peers will use the platform. The build system and the tools
have to support fetching sources from multiple repositories. Why? The
same way you probably could not convince Linus or Keith Packard to
switch to anything other than Git, you probably can't convince other
potencial contributors to switch to Monotone either. And they might
have perfectly good reasons: no version control system is perfect.
The previous 2 sections don't contradict each other. I do see as a
goal to minimize code duplication. But it has to be seen that
OpenEmbedded is mainly a build system + build metadata and it should
not force a version control system on anyone.
PS: I don't want to start the holy VCS wars here. I pretty much think
that all VCSes are bad in one respect or another....and I have used
many of them.
More information about the openmoko-devel