rebaseline_server: allow client to pull all results, or just failures

rebaseline_server: allow client to pull all results, or just failures
(SkipBuildbotRuns)

This will be handy for constrained networks or devices, where we don't want
to bother downloading info about all the successful tests.

R=jcgregorio@google.com

Committed: https://code.google.com/p/skia/source/detail?r=11737

[epoger, 2013-10-10 19:16:07]

Ready for review at patchset 3.

You can try it out for yourself at
http://wpgntat-ubiq141.hot.corp.google.com:8888 , which redirects to
http://wpgntat-ubiq141.hot.corp.google.com:8888/static/view.html?resultsToLoa...
...

You can also try the faster-to-load version at
http://wpgntat-ubiq141.hot.corp.google.com:8888/static/view.html?resultsToLoa...

https://codereview.chromium.org/26891003/diff/3001/gm/rebaseline_server/resul...
File gm/rebaseline_server/results.py (left):

https://codereview.chromium.org/26891003/diff/3001/gm/rebaseline_server/resul...
gm/rebaseline_server/results.py:79: 'index': 0,   # index of this result within
testData list
We generate the "index" entries on the client side now, for 2 reasons:

1. it's not clear whether the indexes should match in the 'all' and 'failures'
lists
2. it seems wasteful to send all these indexes across the wire, when the client
can generate them

https://codereview.chromium.org/26891003/diff/3001/gm/rebaseline_server/resul...
gm/rebaseline_server/results.py:96: def _GetDictsFromRoot(root,
pattern='*.json'):
Updated function/method names as per
http://google-styleguide.googlecode.com/svn/trunk/pyguide.html#Naming

[jcgregorio, 2013-10-10 19:59:10]

https://codereview.chromium.org/26891003/diff/5001/gm/rebaseline_server/serve...
File gm/rebaseline_server/server.py (right):

https://codereview.chromium.org/26891003/diff/5001/gm/rebaseline_server/serve...
gm/rebaseline_server/server.py:157: print 'do_GET_results: sending results of
type "%s"' % type
Can we use the logging module instead of print?

https://codereview.chromium.org/26891003/diff/5001/gm/rebaseline_server/serve...
gm/rebaseline_server/server.py:186: i = string.find(path, '?')
path = urlparse.urlparse(path).path

[epoger, 2013-10-11 17:44:49]

Good suggestions, thanks!  Please see patchset 4...

https://codereview.chromium.org/26891003/diff/5001/gm/rebaseline_server/serve...
File gm/rebaseline_server/server.py (right):

https://codereview.chromium.org/26891003/diff/5001/gm/rebaseline_server/serve...
gm/rebaseline_server/server.py:157: print 'do_GET_results: sending results of
type "%s"' % type
On 2013/10/10 19:59:11, jcgregorio wrote:
> Can we use the logging module instead of print?

Done.

https://codereview.chromium.org/26891003/diff/5001/gm/rebaseline_server/serve...
gm/rebaseline_server/server.py:186: i = string.find(path, '?')
On 2013/10/10 19:59:11, jcgregorio wrote:
> path = urlparse.urlparse(path).path

done.done

[jcgregorio, 2013-10-11 18:15:15]

https://codereview.chromium.org/26891003/diff/9001/gm/rebaseline_server/serve...
File gm/rebaseline_server/server.py (right):

https://codereview.chromium.org/26891003/diff/9001/gm/rebaseline_server/serve...
gm/rebaseline_server/server.py:273: global _SERVER
The global really bothers me, why doesn't Server subclass from HTTPServer, then
it would be available to the handler via the .server instance variable of
handler?

This doesn't have to be fixed in this CL, it could go in another CL.

[epoger, 2013-10-11 18:37:20]

https://codereview.chromium.org/26891003/diff/9001/gm/rebaseline_server/serve...
File gm/rebaseline_server/server.py (right):

https://codereview.chromium.org/26891003/diff/9001/gm/rebaseline_server/serve...
gm/rebaseline_server/server.py:273: global _SERVER
On 2013/10/11 18:15:15, jcgregorio wrote:
> The global really bothers me, why doesn't Server subclass from HTTPServer,
then
> it would be available to the handler via the .server instance variable of
> handler?

Good point, thanks.  Added a TODO above.

[jcgregorio, 2013-10-11 18:38:45]

On 2013/10/11 18:37:20, epoger wrote:
>
https://codereview.chromium.org/26891003/diff/9001/gm/rebaseline_server/serve...
> File gm/rebaseline_server/server.py (right):
> 
>
https://codereview.chromium.org/26891003/diff/9001/gm/rebaseline_server/serve...
> gm/rebaseline_server/server.py:273: global _SERVER
> On 2013/10/11 18:15:15, jcgregorio wrote:
> > The global really bothers me, why doesn't Server subclass from HTTPServer,
> then
> > it would be available to the handler via the .server instance variable of
> > handler?
> 
> Good point, thanks.  Added a TODO above.

LGTM

[epoger, 2013-10-11 18:45:37]

Committed patchset #5 manually as r11737 (presubmit successful).