Use src-side launcher for telemetry_unittests in chromium_trybot recipe

Use src-side launcher for telemetry_unittests in chromium_trybot recipe

Depends on https://codereview.chromium.org/649683005/

BUG=422235
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=292589

[iannucci, 2014-10-21 18:02:46]

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

[iannucci, 2014-10-21 18:02:48]

mostly lgtm, though buildnumber should be deprecated asap

https://codereview.chromium.org/666023004/diff/1/scripts/slave/recipe_modules...
File scripts/slave/recipe_modules/chromium/api.py (right):

https://codereview.chromium.org/666023004/diff/1/scripts/slave/recipe_modules...
scripts/slave/recipe_modules/chromium/api.py:638: for name in ('buildername',
'slavename', 'buildnumber'):
yikes! buildnumber is /extremely dangerous/. What is using it?

[Paweł Hajdan Jr., 2014-10-22 11:18:26]

Thank you for review.

https://codereview.chromium.org/666023004/diff/1/scripts/slave/recipe_modules...
File scripts/slave/recipe_modules/chromium/api.py (right):

https://codereview.chromium.org/666023004/diff/1/scripts/slave/recipe_modules...
scripts/slave/recipe_modules/chromium/api.py:638: for name in ('buildername',
'slavename', 'buildnumber'):
On 2014/10/21 18:02:47, iannucci wrote:
> yikes! buildnumber is /extremely dangerous/. What is using it?

e.g. runtest.py

Could you explain more though why it's dangerous?

By the way I added a TODO to remove this. Note that I believe this is not any
worse than the current state of things where we use buildnumber anyway with
runtest.py. Here we at least have a much shorter list of build properties.

[Paweł Hajdan Jr., 2014-10-22 11:18:43]

The CQ bit was checked by phajdan.jr@chromium.org

[commit-bot: I haz the power, 2014-10-22 11:19:40]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/666023004/20001

[commit-bot: I haz the power, 2014-10-22 11:21:30]

Committed patchset #2 (id:20001) as 292589

[jam, 2014-10-22 16:29:41]

A revert of this CL (patchset #2 id:20001) has been created in
https://codereview.chromium.org/666563007/ by jam@chromium.org.

The reason for reverting is: windows trybots broken

please be around after landing changes and verify that bots are working.

[Paweł Hajdan Jr., 2014-10-22 16:29:52]

A revert of this CL (patchset #2 id:20001) has been created in
https://codereview.chromium.org/675523002/ by phajdan.jr@chromium.org.

The reason for reverting is: still breaking :-/.

[iannucci, 2014-10-22 18:58:42]

https://codereview.chromium.org/666023004/diff/1/scripts/slave/recipe_modules...
File scripts/slave/recipe_modules/chromium/api.py (right):

https://codereview.chromium.org/666023004/diff/1/scripts/slave/recipe_modules...
scripts/slave/recipe_modules/chromium/api.py:638: for name in ('buildername',
'slavename', 'buildnumber'):
On 2014/10/22 11:18:26, Paweł Hajdan Jr. wrote:
> On 2014/10/21 18:02:47, iannucci wrote:
> > yikes! buildnumber is /extremely dangerous/. What is using it?
> 
> e.g. runtest.py
> 
> Could you explain more though why it's dangerous?
> 

It's ephemeral data. It's not tied to anything real, and only exists in the
buildbot database:
  * if you remove and re-add a builder, it resets
  * if you rename the builder, it resets
  * if the master crashes (sometimes we need to wipe old data when that
happens), it can reset
  * if we split the master, it resets
  * if we merge the masters, it resets
  * if the db corrupts (no backups :), it resets
  * if you build out-of-order with the source (e.g. force build), and you use it
to indicate a timeline of source, you'll have out-of-order data points!
  * it is tied to buildbot... there's no way to reproduce a
buildnumber-like-number without a centralized master like buildbot. if we start
getting rid of it now, it'll make our post-buildbot transition that much easier
:)

It's basically not trustworthy for what people expect it to be. A much better
thing to use would be the real revision, or some numerical representation of
that (probably the commit position number).

> By the way I added a TODO to remove this. Note that I believe this is not any
> worse than the current state of things where we use buildnumber anyway with
> runtest.py. Here we at least have a much shorter list of build properties.

Yes, you're correct, it's better than right now :)