Re: Linux 2.6.12-rc3

From: Greg KH
Date: Sun Apr 24 2005 - 00:47:03 EST

On Sun, Apr 24, 2005 at 01:00:23AM +0200, Pavel Machek wrote:
> Hi!
> > > Could we add some kind off "This-changeset-obsoletes: <sha1>" header?
> > > That would allow me to send patches by hand and still make the SCM do the
> > > right thing during merge.
> >
> > That doesn't really scale, plus I don't want to rely on that kind of hack
> > since it's simply not reliable (the patch may have gotten edited on the
> > way, so maybe the stuff I apply is 90% from your patch, but 10%
> > different).
> (Well, at that point I probably want to drop that 10% anyway :-).
> > Also, it doesn't actually handle the generic case, which is that the other
> > end used something else than git to maintain his patches (which in the end
> > has the exact same issues).
> Actually this one should not be a problem. "This-changeset-obsoletes:"
> would probably be in changelog part, and remote end would just
> propagate it.
> > > Alternatively I should just get public rsync-able space somewhere...
> > > Would be willing to add people/pavel?
> >
> > Now, that's actually something people are working on (""),
> > so I don't think that would be a problem. People _are_ trying to set up
> > things like a at least for the kernel. I know OSDL and OSL
> > ( are interested, and I think the current
> > works too.
> >
> > A word of warning: in many ways it's easier to work with patches. In
> > particular, if you want to have me merge from your tree, I require a
> > certain amount of cleanliness in the trees I'm pulling from. All of the
> > people who used to use BK to sync are already used to that, but for people
> > who didn't historically use BK this is going to be a learning experience.
> >
> > The reason patches are easier is that you can start out from a messy tree,
> > and then whittle down the patch to just the part you want to send me, so
> > it doesn't actually matter how messy your original tree is, you can always
> > make the end result look nice.
> I created three trees here (with git fork): one ("clean-git") to track
> your changes, second ("linux-git") to do my development on and third
> ("linux-good") for good, nice, cleaned-up changes, for you to merge.
> ...unfortunately pasky's git just symlinked object/ directories...
> ...that means that if you pull from me using rsync, you'll get all my
> "development" files, too. Not accessible in any normal way, but still
> there.
> That means that git fork can't be used for "good tree for
> Linus"... not until we have something better than rsync :-(.

I'm not really using the git-pasky part of git yet for development
(except for 'git log -c') You can just "clone" the tree yourself with a
stupid little script like I do below. It still uses hard-links so
common git objects are only in one place on the disk.

Feel free to make it better, I have never claimed to be a bash
programmer :)


greg k-h




mkdir $DIR
cd $DIR
mkdir .git
cp ~/linux/ .git/
cp -rl ~/linux/ .git/objects/

HEAD=`cat .git/HEAD`
echo "HEAD=$HEAD"
echo ""

cat-file commit $HEAD
TREE_HEAD=`cat-file commit $HEAD | head -n 1 | cut -f 2 -d " "`

echo "read-tree $TREE_HEAD"
read-tree $TREE_HEAD

echo "checkout-cache -a"
checkout-cache -a

echo "update-cache --refresh"
update-cache --refresh
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at