Restore under flag historic behaviour of lastchange.py wrt. Git-SVN repos

Restore under flag historic behaviour of lastchange.py wrt. Git-SVN repos

Adds optional flag '--git-svn-go-deeper' to allow lastchange.py find
the last committed SVN revision in a Git-SVN repo.

This is to help access DevTools frontend on Android for Chromium developers
that have local changes to Blink.

BUG=468691
Committed: https://crrev.com/b73e2ea7a4e319eb2e6183090dfe6b9796d42c3b
Cr-Commit-Position: refs/heads/master@{#335156}

[mnaganov (inactive), 2015-06-18 21:57:39]

mnaganov@chromium.org changed reviewers:
+ scottmg@chromium.org

[mnaganov (inactive), 2015-06-18 21:57:39]

Hi Scott,

PTAL. This is to work around the effects of this change:
https://codereview.chromium.org/488733002/. I totally understand what it was
needed, but it started to create problems for people who have local Blink
changes, as lastchange.py may now override LASTCHANGE.blink with a local git
commit hash that doesn't make any sense to DevTools Frontend server (see
http://crbug.com/468691).

The long-term solution will be to start using Cr-Commit-Position once Blink repo
merges with Chromium, but as this isn't happening quickly, the short-term
solution that I see is as follows:

  -- provide this flag to lastchange.py that allows generating correct
LASTCHANGE.blink by hand;
  -- provide a customized 404 page on the DevTools Frontend server that will
hint people to run this command locally and rebuild.

I see attempts to retrieve DevTools frontend using a git hash almost every day,
so the problem is real.

[scottmg, 2015-06-18 22:17:14]

On 2015/06/18 21:57:39, mnaganov wrote:
> Hi Scott,
> 
> PTAL. This is to work around the effects of this change:
> https://codereview.chromium.org/488733002/. I totally understand what it was
> needed, but it started to create problems for people who have local Blink
> changes, as lastchange.py may now override LASTCHANGE.blink with a local git
> commit hash that doesn't make any sense to DevTools Frontend server (see
> http://crbug.com/468691).
> 
> The long-term solution will be to start using Cr-Commit-Position once Blink
repo
> merges with Chromium, but as this isn't happening quickly, the short-term
> solution that I see is as follows:
> 
>   -- provide this flag to lastchange.py that allows generating correct
> LASTCHANGE.blink by hand;
>   -- provide a customized 404 page on the DevTools Frontend server that will
> hint people to run this command locally and rebuild.
> 
> I see attempts to retrieve DevTools frontend using a git hash almost every
day,
> so the problem is real.

This seems a bit yucky. Why does it need a version/hash/whatever at all?

[mnaganov (inactive), 2015-06-18 22:38:03]

On 2015/06/18 22:17:14, scottmg wrote:
> On 2015/06/18 21:57:39, mnaganov wrote:
> > Hi Scott,
> > 
> > PTAL. This is to work around the effects of this change:
> > https://codereview.chromium.org/488733002/. I totally understand what it was
> > needed, but it started to create problems for people who have local Blink
> > changes, as lastchange.py may now override LASTCHANGE.blink with a local git
> > commit hash that doesn't make any sense to DevTools Frontend server (see
> > http://crbug.com/468691).
> > 
> > The long-term solution will be to start using Cr-Commit-Position once Blink
> repo
> > merges with Chromium, but as this isn't happening quickly, the short-term
> > solution that I see is as follows:
> > 
> >   -- provide this flag to lastchange.py that allows generating correct
> > LASTCHANGE.blink by hand;
> >   -- provide a customized 404 page on the DevTools Frontend server that will
> > hint people to run this command locally and rebuild.
> > 
> > I see attempts to retrieve DevTools frontend using a git hash almost every
> day,
> > so the problem is real.
> 
> This seems a bit yucky. Why does it need a version/hash/whatever at all?

Because DevTools consist of the "host" part that resides in Blink, and the
frontend, which is effectively, a web app. Frontend is not shipped on Android to
reduce Chrome's apk size, instead it is fetched from the server based on Blink
revision used for this particular build of Chrome. It needs to be a Blink
revision, as the communication protocol between the host and the frontend have
private part, which evolves rapidly.

What exactly do you find disturbing in this change?

[scottmg, 2015-06-18 22:55:47]

Nothing disturbing, just yet-another-flag.

Thanks, lgtm then.

[scottmg, 2015-06-18 22:56:35]

https://codereview.chromium.org/1197483002/diff/1/build/util/lastchange.py
File build/util/lastchange.py (right):

https://codereview.chromium.org/1197483002/diff/1/build/util/lastchange.py#ne...
build/util/lastchange.py:285: opts.default_lastchange,
this looks oddly indented, if it's not from an autoformatter, please do +4 or
align at (

[mnaganov (inactive), 2015-06-18 22:59:50]

Thanks so much!

https://codereview.chromium.org/1197483002/diff/1/build/util/lastchange.py
File build/util/lastchange.py (right):

https://codereview.chromium.org/1197483002/diff/1/build/util/lastchange.py#ne...
build/util/lastchange.py:285: opts.default_lastchange,
On 2015/06/18 22:56:34, scottmg wrote:
> this looks oddly indented, if it's not from an autoformatter, please do +4 or
> align at (

Done.

[mnaganov (inactive), 2015-06-18 23:00:02]

The CQ bit was checked by mnaganov@chromium.org

[mnaganov (inactive), 2015-06-18 23:00:03]

The patchset sent to the CQ was uploaded after l-g-t-m from scottmg@chromium.org
Link to the patchset: https://codereview.chromium.org/1197483002/#ps20001
(title: "Comments addressed")

[commit-bot: I haz the power, 2015-06-18 23:01:45]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1197483002/20001

[commit-bot: I haz the power, 2015-06-18 23:16:30]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-06-18 23:17:13]

Patchset 2 (id:??) landed as
https://crrev.com/b73e2ea7a4e319eb2e6183090dfe6b9796d42c3b
Cr-Commit-Position: refs/heads/master@{#335156}