Seeking information about lightweight copying/renaming

[Sune Foldager]

On 06-04-2010 17:24, Paul Malmsten wrote:
> Hi Sune,
> 
> Thank you very much for your insight.
> 
> On Tue, Apr 6, 2010 at 7:58 AM, Sune Foldager <cryo at cyanite.org> wrote:
>> Have there been any decisions on this that I missed? At any rate:
> 
> As far as I know, you haven't missed much of any discussion. I've just
> recently been exploring the potential of picking up where previous
> lightweight copy work left off as a GSoC project.

Right, okay. To be honest, I was hoping to do this work "myself",
although like I said I have so far only managed to rebase the patch
queue, combine a few patches, get an overview, plan the wire stuff and
push a small patch out.

I work for a company where we use Mercurial and we are looking very much
forward to lwcopy as well.

> What were the protocol issues which prevented further progress?

Well, it's some details about how to extend the current protocol in a
more flexible way. It turns out that the current http protocol has more
flexibility in most respects than the ssh one, so it was work to unify
the feature set by introducing a new protocol version.

It stranded a bit due to disagreements over the design, which have yet
to be resolved... and hopefully very soon.

> What is your vision for how work should continue on the project? Would you
> mind sending me a link to your work?

So, my work is mainly a rebased and slightly changed patch queue. I will
at any rate put it on bitbucket soon.

> That's true, although some of the current tests md5sum the revlogs, so if
> they are affected by a lightweight copy feature, they will need to be
> updated. Also, my thought was to add one or more tests which verify correct
> creation/reading of lightweight copy revlogs in isolation, perhaps with a
> similar approach.

Yes... the latter sounds more like unit tests, which we don't have many
of. But that's not a reason for not writing them :). As for the hashes,
they will stay the same, ensuring full compatability between the same
repo running with or without lwcopy.

> Ah, okay, the partial double-implementation threw me off a little. Thanks
> for the clarification.

You're welcome :)

/Sune