Potential way to support git remotes/branches in Rietveld.

Potential way to support git remotes/branches in Rietveld.

[dhermes, 2013-07-02 21:25:47]

Andi,

I didn't send this to the larger list because I'm not sure I've done enough/not
sure it's going in the right direction.

So this is more of a "discuss my stuff" instead of "review my stuff".

Let me know what you think. I feel like this is a hairy mess to wade through and
I've only gotten to the tip of the iceberg.

[Andi, 2013-07-08 03:42:10]

Hi Danny,

sorry for the delay!

Such a feature was requested several times. After looking through your CS I have
some questions:

- Wouldn't be git_remote and git_remote_branch useful for Mercurial too? When
using Mercurial we could use the same fields to identify repos and branches too.
Or do I miss something, because you explicitly distinct between git and
git/mercurial in the models?

- We should identify remote_url, remote_branch and rev automatically in
upload.py. That should be possible with bot git and Mercurial. We could leave
the command line options to override them, but it would be much more comfortable
to populate them always.

[dhermes, 2013-07-15 21:37:40]

I have updated the change with renaming the variables.

RE: using relative_commit_hash and --rev interchangeable:

This is somewhat problematic because people could use HEAD, master,
<branch-name>, abbreviated hash, etc. whereas I have the field be 40 chars
(I'm now realizing I should make it be 40 hex chars).

This could be resolved by taking the --rev value and passing to rev-parse:

bossylobster:git-rv (master) dhermes$ git rev-parse HEAD
7180948d11683cd3f1d901393e8328e6064a5b50
bossylobster:git-rv (master) dhermes$ git rev-parse 71809
7180948d11683cd3f1d901393e8328e6064a5b50
bossylobster:git-rv (master) dhermes$ git rev-parse master
7180948d11683cd3f1d901393e8328e6064a5b50



On Sun, Jul 7, 2013 at 8:42 PM, <albrecht.andi@gmail.com> wrote:

> Hi Danny,
>
> sorry for the delay!
>
> Such a feature was requested several times. After looking through your
> CS I have some questions:
>
> - Wouldn't be git_remote and git_remote_branch useful for Mercurial too?
> When using Mercurial we could use the same fields to identify repos and
> branches too. Or do I miss something, because you explicitly distinct
> between git and git/mercurial in the models?
>
> - We should identify remote_url, remote_branch and rev automatically in
> upload.py. That should be possible with bot git and Mercurial. We could
> leave the command line options to override them, but it would be much
> more comfortable to populate them always.
>
>
https://codereview.appspot.**com/10891043/<https://codereview.appspot.com/108...
>



-- 
Danny Hermes
Developer Programs Engineer

[bossylobster, 2013-08-19 21:34:52]

Bump

[Andi, 2013-08-26 09:29:12]

LGTM

Do you have a test instance running somewhere?