Re: Re: [PATCH] [submodule] Add depth to submodule update

[Heiko Voigt <hvoigt@hvoigt.net>]

On Thu, Jun 27, 2013 at 04:54:45PM +0200, Jens Lehmann wrote:
> Am 26.06.2013 23:03, schrieb Junio C Hamano:
> > Fredrik Gustafsson <iveqy@iveqy.com> writes:
> > 
> >> On Wed, Jun 26, 2013 at 12:11:32AM +0200, Heiko Voigt wrote:
> >>> On Tue, Jun 25, 2013 at 12:49:25AM +0200, Fredrik Gustafsson wrote:
> >>>> Used only when a clone is initialized. This is useful when the submodule(s)
> >>>> are huge and you're not really interested in anything but the latest commit.
> >>>>
> >>>> Signed-off-by: Fredrik Gustafsson <iveqy@iveqy.com>
> >>>
> >>> I this is a valid use case. But this option only makes sense when a
> >>> submodule is newly cloned so I am not sure whether submodule update is
> >>> the correct place. Let me think about this a little more. Since we do
> >>> not have any extra command that initiates the clone this is probably the
> >>> only place we can put this option. But at the moment it does not feel
> >>> completely right.
> >>>
> >>> Apart from that the code looks good. If the user does a checkout of a
> >>> revision that was not fetched submodule update will error out the same
> >>> way as if someone forgot to push his submodule changes. So that should
> >>> not be a problem.
> >>
> >> I agree and would love to say that I've a more beautiful solution, but
> >> I haven't.
> >>
> >> The only other solution I can think about is to add a git
> >> submodule clone that will do only clones of non-cloned submodules.
> > 
> > The "update" subcommand already has "--init" to do "init && update",
> > and it would not complain if a given submodule is what you already
> > have shown interest in, so in that sense, I do not think what the
> > posted patch does is too bad---if it is already cloned, it just
> > ignores the depth altogether and makes sure the repository is there.
> > A separate "submodule clone" would only make it more cumbersome to
> > use, I suspect.
> 
> Yup, I see no need for a new command either.

I agree there is no reason for that.

> Me too thinks adding "--depth" to "update" makes sense (and I don't
> think that this pretty generic name will become a problem later in
> case someone wants to add a maximum recursion depth, as grep already
> uses "--max-depth" for the same purpose).

Hmm, but does it have a --depth option for revisions? Maybe we should
call it --clone-depth or --rev-depth to make it clear? --depth and
--max-depth would be completely orthogonal but the name does not allow
to distinguish them properly.

> But "--depth" should also be added to the "submodule add" command.
> As an example we already have the "--reference" option, which is
> passed to clone on add and update. Additionally that one supports
> the form with and without '=', so I'd prefer the new update option
> to basically re-use the same code the reference option uses. And
> at least two tests, of course ;-)

And add documentation, please :-)

Cheers Heiko