Use tracking remote name for gerrit upload.

Use tracking remote name for gerrit upload.

This patch replaces the hard-coded remote name 'origin' with the
one obtained by git-remote.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=299491

[alokp, 2016-03-24 21:08:35]

alokp@chromium.org changed reviewers:
+ iannucci@chromium.org

[alokp, 2016-03-24 21:08:36]

Robbie: I do not know the history behind the hard-coded remote name for gerrit
upload, but ran into this limitation while trying to make it work for chromecast
internal repo that uses gerrit.

Is there a particular reason why it is hardcoded to be origin? This patch seems
to work for us but I am not sure if it breaks anything else. PTAL.

[iannucci, 2016-03-24 21:26:44]

iannucci@chromium.org changed reviewers:
+ ukai@chromium.org

[iannucci, 2016-03-24 21:26:44]

It looks like ukai@ committed the original.

Specifically, I know nothing about this function or its assumptions. In general,
depot_tools tools all tend to assume 'origin' at one point or another.

[iannucci, 2016-03-24 21:27:48]

iannucci@chromium.org changed reviewers:
+ tandrii@chromium.org

[iannucci, 2016-03-24 21:27:48]

+tandrii@ who's also been working around the git-cl tools recently

[tandrii(chromium), 2016-03-24 21:49:30]

Thanks for patch! Have you ran the tests? If you did, LGTM. Land through CQ
then.

I'm just slightly surprised this didn't result in a change of expectations in
tests/git_cl_test.py.

Indeed, I'm currently essentially rewriting Gerrit functionality for git cl. See
http://crbug.com/579160. The number of reverts on that bug suggests that
checking what will be broken is hard - test coverage is very insufficient, so
that's maybe why the expectations didn't change.

History-wise, git cl support for Gerrit exists for "git cl upload" only, and
even that is very incomplete. But i'm changing that.

Btw, this CL of mine will likely conflict with yours
https://codereview.chromium.org/1830973003, but I think you can go ahead. I'll
patch mine afterwards and land tomorrow, as I'm too tired at the moment.

[alokp, 2016-03-24 23:41:39]

Thanks for your review. I did not run the tests. I am not sure which one or how
to run them. git-cl upload did gave me some warnings about expectations but I am
not sure what to do there. If you point me to any documentation I am happy to
make the tests/expectations pass.

[ukai, 2016-03-25 01:47:32]

lgtm

[alokp, 2016-03-25 04:33:02]

I am going to try the CQ to see if any expectations need updating.

[alokp, 2016-03-25 04:33:03]

I am going to try the CQ to see if any expectations need updating.

[alokp, 2016-03-25 04:33:07]

The CQ bit was checked by alokp@chromium.org

[commit-bot: I haz the power, 2016-03-25 04:33:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1830313002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1830313002/1

[commit-bot: I haz the power, 2016-03-25 04:35:54]

Description was changed from

==========
Use tracking remote name for gerrit upload.

This patch replaces the hard-coded remote name 'origin' with the
one obtained by git-remote.
==========

to

==========
Use tracking remote name for gerrit upload.

This patch replaces the hard-coded remote name 'origin' with the
one obtained by git-remote.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=299491
==========

[commit-bot: I haz the power, 2016-03-25 04:35:55]

Committed patchset #1 (id:1) as
http://src.chromium.org/viewvc/chrome?view=rev&revision=299491

[tandrii(chromium), 2016-03-25 17:03:22]

oh, yet another example of insufficient test coverage :(

[tandrii(chromium), 2016-03-28 15:41:45]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/1835923002/ by tandrii@chromium.org.

The reason for reverting is: Argh, this has broken uploading track-ed branches.
Repro:

git new-branch base-feature
touch x1 && git add x1 && git commit -m "x1"
git cl upload
git new-branch --upstream_current dep-feature
touch y1 && git add y1 && git commit -m "y1"
git cl upload -v -v # fails.


the verbose output produces this:
---cut---
DEBUG:root:git push .
7079b1682431184375b62e7e81ce10a02171aa0d:refs/for/refs/heads/master
To .
 ! [rejected]        7079b1682431184375b62e7e81ce10a02171aa0d ->
refs/for/refs/heads/master (non-fast-forward)
error: failed to push some refs to '.'
---cut---.