build: Bot_update refactor recipe roll

build: Bot_update refactor recipe roll

Manual Recipe Roll for https://codereview.chromium.org/1686273002/
Includes a couple of changes that bypasses bot_update checking for did_run

BUG=

[Ryan Tseng, 2016-02-16 18:55:22]

hinoka@google.com changed reviewers:
+ estaab@chromium.org, hinoka@google.com, iannucci@chromium.org

[estaab, 2016-02-16 20:19:51]

estaab@chromium.org changed reviewers:
- estaab@chromium.org

[estaab, 2016-02-16 20:19:51]

estaab@chromium.org changed required reviewers:
+ iannucci@chromium.org

[iannucci, 2016-02-16 21:04:53]

I want to verify that this has no actual semantic change: no bots that were
previously NOT using bot_update will suddenly start using it, right?

https://chromiumcodereview.appspot.com/1694383003/diff/1/scripts/slave/recipe...
File scripts/slave/recipe_modules/chromium_android/api.py (left):

https://chromiumcodereview.appspot.com/1694383003/diff/1/scripts/slave/recipe...
scripts/slave/recipe_modules/chromium_android/api.py:119: result =
self.m.gclient.checkout(spec, with_branch_heads=with_branch_heads)
Did chromium_android use bot_update before? Or was it blacklisted out?

https://chromiumcodereview.appspot.com/1694383003/diff/1/scripts/slave/recipe...
File scripts/slave/recipes/swarming/staging.py (right):

https://chromiumcodereview.appspot.com/1694383003/diff/1/scripts/slave/recipe...
scripts/slave/recipes/swarming/staging.py:80: step_result =
api.bot_update.ensure_checkout()
this was previously not force=True, are we sure it was being used?

[Ryan Tseng, 2016-02-16 22:09:03]

So there were a couple that (according to expectations) were on gclient and got
inadvertently switched to bot_update :/.

I fixed the two cases i found (a couple in chormium_android, and also the
swarming builder).  There's one more builder (the webrtc ios) that was inactive
in the expectations, but there was no gclient fallback, so i'm convinced that
the expectations were wrong before.

If the expectations are to be trusted, then yes, there is no semantic change
(other than webrtc ios).

[estaab, 2016-02-16 22:25:57]

I think you can run this through "Build Try Recipe Test" on tryserver.infra
since tandrii added it in:
https://codereview.chromium.org/1664543002/

[iannucci, 2016-02-16 23:55:04]

Hm... This is still unexpectedly huge.

Can we revert the upstream change and re-land this one change-at-a-time?

  * build: remove dependency on did_run
  * depot_tools: remove master/builder/slavename (and relevant changes to
bot_update.py)
  * depot_tools: remove did_run boolean
  * depot_tools: remove deprecated command line arguments

[iannucci, 2016-02-16 23:57:27]

probably also need to make sure build_internal doesn't depend on did_run.

[hinoka, 2016-02-17 00:04:35]

So basically start over?

Also I think you listed some of the least risky portion of the change. I'm
more worried about unexpected behavior from removing deps2git,
build_internal, and svn revisions.

On Tue, Feb 16, 2016 at 3:55 PM, <iannucci@chromium.org> wrote:

> Hm... This is still unexpectedly huge.
>
> Can we revert the upstream change and re-land this one change-at-a-time?
>
>   * build: remove dependency on did_run
>   * depot_tools: remove master/builder/slavename (and relevant changes to
> bot_update.py)
>   * depot_tools: remove did_run boolean
>   * depot_tools: remove deprecated command line arguments
> https://chromiumcodereview.appspot.com/1694383003/
>
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[iannucci, 2016-02-17 00:19:20]

On 2016/02/17 00:04:35, hinoka wrote:
> So basically start over?
> 

Well take what you have, split and land incrementally.

I didn't enumerate all the dangerous things :)

The goal is to identify small, relatively safe changes, and land those
individually.  For dangerous changes, the downstream expectations should
hopefully be changed ahead of time (like the did_run thing).

[hinoka, 2016-02-17 00:21:47]

Ok we can do this tomorrow when you're around.  This is gonna result in
quite a number of CLs

On Tue, Feb 16, 2016 at 3:57 PM, <iannucci@chromium.org> wrote:

> probably also need to make sure build_internal doesn't depend on did_run.
> https://chromiumcodereview.appspot.com/1694383003/
>
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[iannucci, 2016-02-17 00:25:41]

OK