Added support for re-running of layout tests form the dashboard.

Adding support for re-running of layout tests form the dashboard.

This allows the developer the ease of selecting specific layout tests which they want to re run from the convinience of a web page instead of terminal.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=182377

[vivekg, 2014-08-26 11:05:13]

Please add some unit tests for supported actions, running port etc. for the new
functionality.

[patro, 2014-08-26 11:11:25]

Patchset #1 (id:1) has been deleted

[patro, 2014-08-26 15:02:46]

behara.ms@samsung.com changed reviewers:
+ dpranke@chromium.org

[patro, 2014-09-01 14:25:36]

Kindly Review

Thanks,
Shyam Patro

[Dirk Pranke, 2014-09-03 00:01:31]

Using websockets for this seems kinda goofy, particularly since this is
basically a single RPC and doesn't really need bidirectional messaging over a
socket.

When we've done this sort of thing in the past we've built simple http servers
on top of BaseHttpServer. There is a scaffolding class for this in
webkitpy/tool/servers/reflectionserver.py .

That was how we ran garden-o-matic locally before we moved to using an AppEngine
instance. 

I encourage you to take a look at that instead of building an entirely new
client/server mechanism. 

If there's some reasons we should be using websockets though, please let me
know.

[patro, 2014-09-10 12:55:04]

On 2014/09/03 00:01:31, Dirk Pranke wrote:
> Using websockets for this seems kinda goofy, particularly since this is
> basically a single RPC and doesn't really need bidirectional messaging over a
> socket.
> 
> When we've done this sort of thing in the past we've built simple http servers
> on top of BaseHttpServer. There is a scaffolding class for this in
> webkitpy/tool/servers/reflectionserver.py .
> 
> That was how we ran garden-o-matic locally before we moved to using an
AppEngine
> instance. 
> 
> I encourage you to take a look at that instead of building an entirely new
> client/server mechanism. 
> 
> If there's some reasons we should be using websockets though, please let me
> know.

Yes this can be done using only single RPC doesnt't need bi-directional
messaging.

But some kind of feedback from the server may be needed if the tests are
running/failed/succesfully completed.

And further more it will be more interactive if we can display the output of the
script on the webpage so that user gets to know what is happening in the
background.

[Dirk Pranke, 2014-09-10 20:17:31]

On 2014/09/10 12:55:04, patro wrote:
> On 2014/09/03 00:01:31, Dirk Pranke wrote:
> > Using websockets for this seems kinda goofy, particularly since this is
> > basically a single RPC and doesn't really need bidirectional messaging over
a
> > socket.
> > 
> > When we've done this sort of thing in the past we've built simple http
servers
> > on top of BaseHttpServer. There is a scaffolding class for this in
> > webkitpy/tool/servers/reflectionserver.py .
> > 
> > That was how we ran garden-o-matic locally before we moved to using an
> AppEngine
> > instance. 
> > 
> > I encourage you to take a look at that instead of building an entirely new
> > client/server mechanism. 
> > 
> > If there's some reasons we should be using websockets though, please let me
> > know.
> 
> Yes this can be done using only single RPC doesnt't need bi-directional
> messaging.
> 
> But some kind of feedback from the server may be needed if the tests are
> running/failed/succesfully completed.
> 
> And further more it will be more interactive if we can display the output of
the
> script on the webpage so that user gets to know what is happening in the
> background.

You can do the latter via  a long running POST that streams the output from
run-webkit-tests back to the browser (this is essentially what buildbot does). 

You could even do something more responsive where you were parsing the output as
it ran and streamed updates back as part of the response (though the output from
run-webkit-tests currently isn't very parser-friendly).

I still don't see websockets as a compelling design here.

[patro, 2014-09-11 12:22:11]

behara.ms@samsung.com changed reviewers:
+ kphanee@chromium.org

[patro, 2014-09-12 14:29:05]

Patchset #4 (id:80001) has been deleted

[patro, 2014-09-12 14:34:44]

On 2014/09/10 20:17:31, Dirk Pranke wrote:
> On 2014/09/10 12:55:04, patro wrote:
> > On 2014/09/03 00:01:31, Dirk Pranke wrote:
> > > Using websockets for this seems kinda goofy, particularly since this is
> > > basically a single RPC and doesn't really need bidirectional messaging
over
> a
> > > socket.
> > > 
> > > When we've done this sort of thing in the past we've built simple http
> servers
> > > on top of BaseHttpServer. There is a scaffolding class for this in
> > > webkitpy/tool/servers/reflectionserver.py .
> > > 
> > > That was how we ran garden-o-matic locally before we moved to using an
> > AppEngine
> > > instance. 
> > > 
> > > I encourage you to take a look at that instead of building an entirely new
> > > client/server mechanism. 
> > > 
> > > If there's some reasons we should be using websockets though, please let
me
> > > know.
> > 
> > Yes this can be done using only single RPC doesnt't need bi-directional
> > messaging.
> > 
> > But some kind of feedback from the server may be needed if the tests are
> > running/failed/succesfully completed.
> > 
> > And further more it will be more interactive if we can display the output of
> the
> > script on the webpage so that user gets to know what is happening in the
> > background.
> 
> You can do the latter via  a long running POST that streams the output from
> run-webkit-tests back to the browser (this is essentially what buildbot does).

> 
> You could even do something more responsive where you were parsing the output
as
> it ran and streamed updates back as part of the response (though the output
from
> run-webkit-tests currently isn't very parser-friendly).
> 
> I still don't see websockets as a compelling design here.

You are right BaseHttpServer is simple and is adequate for the purpose.
Modified previous patch to use BaseHttpServer instead of websockets.

Steps to Use:
1)Open dashboard.html from a previous run of run-webkit-tests
layout-test-results folder
2)Select the tests you want to run using the checkboxes
3)Click Re-run
4)The tests selected are run and the live output form run-webkit-tests is posted
onto the webpage.

Kindly Review,

Thanks
Shyam Patro

[Dirk Pranke, 2014-09-12 18:43:05]

okay, this looks generally close to right.

For consistency with the other servers, though, it'd be better if we moved this
to webkitpy/tool/servers and created a webkit-patch command to wrap it (see
tool/servers/rebaselineserver.py and tool/commands/rebaselineserver.py for an
example).

[patro, 2014-09-19 10:19:45]

Patchset #5 (id:120001) has been deleted

[patro, 2014-09-19 10:32:51]

On 2014/09/12 18:43:05, Dirk Pranke wrote:
> okay, this looks generally close to right.
> 
> For consistency with the other servers, though, it'd be better if we moved
this
> to webkitpy/tool/servers and created a webkit-patch command to wrap it (see
> tool/servers/rebaselineserver.py and tool/commands/rebaselineserver.py for an
> example).

Done adding the webkit-patch command wrapper 

Kindly Review,

Shyam Patro

[Dirk Pranke, 2014-09-19 18:34:26]

lgtm. Thanks for bearing with me!

https://codereview.chromium.org/502953003/diff/140001/Tools/Scripts/webkitpy/...
File Tools/Scripts/webkitpy/tool/servers/layouttestsserver.py (right):

https://codereview.chromium.org/502953003/diff/140001/Tools/Scripts/webkitpy/...
Tools/Scripts/webkitpy/tool/servers/layouttestsserver.py:51: # Fixme : Has to be
json object
nit: we tend to use FIXME: in all caps.

[patro, 2014-09-20 17:11:31]

https://codereview.chromium.org/502953003/diff/140001/Tools/Scripts/webkitpy/...
File Tools/Scripts/webkitpy/tool/servers/layouttestsserver.py (right):

https://codereview.chromium.org/502953003/diff/140001/Tools/Scripts/webkitpy/...
Tools/Scripts/webkitpy/tool/servers/layouttestsserver.py:51: # Fixme : Has to be
json object
On 2014/09/19 18:34:26, Dirk Pranke wrote:
> nit: we tend to use FIXME: in all caps.

Done.

[patro, 2014-09-20 17:11:42]

The CQ bit was checked by behara.ms@samsung.com

[commit-bot: I haz the power, 2014-09-20 17:11:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/502953003/160001

[commit-bot: I haz the power, 2014-09-20 18:11:15]

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

[commit-bot: I haz the power, 2014-09-20 18:11:16]

Try jobs failed on following builders:
  linux_blink_rel on tryserver.blink
(http://build.chromium.org/p/tryserver.blink/builders/linux_blink_rel/builds/2...)

[patro, 2014-09-21 18:41:31]

The CQ bit was checked by behara.ms@samsung.com

[commit-bot: I haz the power, 2014-09-21 18:42:55]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/502953003/180001

[commit-bot: I haz the power, 2014-09-21 19:16:53]

Committed patchset #7 (id:180001) as 182377