On Mon, 2010-01-25 at 11:53 +0000, Martyn Welch wrote:
> Richard Purdie wrote:
> > On Mon, 2010-01-25 at 10:43 +0000, Martyn Welch wrote:
> >
> >> The Git fetcher currently hardwires the git command to "git". Allow the
> >> path and any additional wrappers to the Git command to be provided via
> >> FETCHCMD functionality, as with some of the other fetchers.
> >>
> >> This requires FETCHCMD_svn to be defined in bitbake.conf.
> >>
> >
> > I have to ask why? I suspect this is so you can deal with firewalls? If
> > thats the case I'd suggest you look at .gitconfig files instead as
> > they're powerful and don't need custom wrappers to git.
> >
> > Poky has a function, generate_git_config(e) in
> > http://git.pokylinux.org/cgit.cgi/poky/tree/meta/classes/base.bbclass
> > which allows construction of a .gitconfig file from two variables,
> > GIT_PROXY_IGNORE and GIT_PROXY_COMMAND.
> >
> > This means you can have servers on both the sides of the firewall and
> > git can fetch from both
> Yes, it's for firewalls. We have the Dante client already set up, which
> if I understand correctly provides an equivalent to the method you
> suggest. The advantage we see in this approach is that the Dante client
> can be used for all the fetchers and is not specific to git.
There is no harm making it configurable I guess. You need to do
something like
basecmd = data.getVar("FETCHCMD_git", d, True) or "git"
though to ensure a good default.
Also, the ls-remote call is in a different function and will most likely
error with your patch as basecmd won't exist. I'd set ud.basecmd = ...
in the localpath function and then use that...
Cheers,
Richard
_______________________________________________
Bitbake-dev mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/bitbake-dev
