chromium_tests: raise infra failure when no compile targets are specified

chromium_tests: raise infra failure when no compile targets are specified

This relies on default compile targets either on the recipe side or ninja,
but breaks badly with e.g. analyze. We're trying to make as much of bot
config explicit as possible.

BUG=536186
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=296903

[Paweł Hajdan Jr., 2015-09-28 10:46:21]

phajdan.jr@chromium.org changed reviewers:
+ sergiyb@chromium.org, tandrii@chromium.org

[Paweł Hajdan Jr., 2015-09-28 10:46:39]

The CQ bit was checked by phajdan.jr@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-09-28 10:46:54]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1376453002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1376453002/1

[commit-bot: I haz the power, 2015-09-28 10:49:52]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-09-28 10:49:53]

Dry run: This issue passed the CQ dry run.

[Sergiy Byelozyorov, 2015-09-28 13:07:47]

Are we expecting a lot of bots that rely on default compile targets? If yes,
this would rapidly increase number of infra failures and may result in
significant increase in false rejection rate. Perhaps we should first issue a
warning (orange step) rather than failing it and then use monitoring to
determine which builders rely on default compile targets.

[Paweł Hajdan Jr., 2015-09-28 13:09:45]

On 2015/09/28 at 13:07:47, sergiyb wrote:
> Are we expecting a lot of bots that rely on default compile targets? If yes,
this would rapidly increase number of infra failures and may result in
significant increase in false rejection rate. Perhaps we should first issue a
warning (orange step) rather than failing it and then use monitoring to
determine which builders rely on default compile targets.

How will we find the orange steps?

I'd much rather just land this and fix the issues.

Currently we're having known false negatives because of this, which IMO is a
worse failure mode.

[Sergiy Byelozyorov, 2015-09-28 13:38:28]

On 2015/09/28 13:09:45, Paweł Hajdan Jr. wrote:
> On 2015/09/28 at 13:07:47, sergiyb wrote:
> > Are we expecting a lot of bots that rely on default compile targets? If yes,
> this would rapidly increase number of infra failures and may result in
> significant increase in false rejection rate. Perhaps we should first issue a
> warning (orange step) rather than failing it and then use monitoring to
> determine which builders rely on default compile targets.
> 
> How will we find the orange steps?
> 
> I'd much rather just land this and fix the issues.
> 
> Currently we're having known false negatives because of this, which IMO is a
> worse failure mode.

Agreed. lgtm. Regarding how to find, I can explain offline.

[Paweł Hajdan Jr., 2015-09-28 13:40:10]

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

[commit-bot: I haz the power, 2015-09-28 13:40:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1376453002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1376453002/1

[commit-bot: I haz the power, 2015-09-28 13:43:30]

Committed patchset #1 (id:1) as
http://src.chromium.org/viewvc/chrome?view=rev&revision=296903

[hans, 2015-09-28 16:50:05]

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

[hans, 2015-09-28 16:50:06]

https://codereview.chromium.org/1376453002/diff/1/scripts/slave/recipes/chrom...
File
scripts/slave/recipes/chromium.expected/full_chromium_fyi_ClangToTLinux.json
(right):

https://codereview.chromium.org/1376453002/diff/1/scripts/slave/recipes/chrom...
scripts/slave/recipes/chromium.expected/full_chromium_fyi_ClangToTLinux.json:160:
"reason": "Misconfigured bot: no compile targets",
Wait, are you deliberately breaking our bots here??

[Nico, 2015-09-28 18:13:58]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/1377453003/ by thakis@chromium.org.

The reason for reverting is: This is not a valid assumption, e.g. the clang bots
all use the empty compile target as build target and were all broken with this.
(As discussed previously, even!).