[Piglit] recovering from aborted runs

[Brian Paul]

On 03/23/2015 03:38 PM, Dylan Baker wrote:
> On Mon, Mar 23, 2015 at 03:24:21PM -0600, Brian Paul wrote:
>> On 03/23/2015 03:07 PM, Dylan Baker wrote:
>>> On Mon, Mar 23, 2015 at 02:57:07PM -0600, Brian Paul wrote:
>>>> Hmm, can you give me a concrete example of how to do this?
>>>>
>>>> I have an aborted run and tests/ contains several thousand NNN.json
>>>> files.  What "piglit summary console" command should I use to generate
>>>> the json file?  I've tried a few things without luck.
>>>
>>> `piglit summary console -s <your results folder>`
>>> gives me a results.json file.
>>
>> OK, if I leave off the 'tests/' part I get the summary info, but I still
>
> You are correct, I happened to start a run into a directory that already
> had a run. I'll send a patch, sorry about the confusion.

OK, thanks.


>
>> don't get a consolidated .json file.  When I have 'tests/' at the end of
>> the path, the script fails:
>>
>> $ bin/piglit.py summary console -s results/03-23/tests/
>> Traceback (most recent call last):
>>     File "bin/piglit.py", line 148, in <module>
>>       main()
>>     File "bin/piglit.py", line 143, in main
>>       returncode = parsed.func(args)
>>     File
>> "/home/Brian/piglit-mingw32/lib/piglit/framework/programs/summary.py",
>> line 136, in console
>>       output = summary.Summary(args.results)
>>     File "/home/Brian/piglit-mingw32/lib/piglit/framework/summary.py",
>> line 301, in __init__
>>       self.results = [framework.results.load_results(i) for i in resultfiles]
>>     File "/home/Brian/piglit-mingw32/lib/piglit/framework/results.py",
>> line 217, in load_results
>>       raise Exception("No results found")
>> Exception: No results found
>
> That is expected.

Hmm, when I see a python traceback I generally assume that something 
unexpected happened.  Wouldn't it be better to catch the exception and 
print a helpful error message?  The above "No results found" didn't help 
me understand the problem.

-Brian

>
>>
>>
>> -Brian
>>
>>>
>>>>
>>>> BTW, the piglit.py script always dumps core upon exit on Cygwin on
>>>> Windows.  I don't know why.
>>>>
>>>> -Brian
>>>> %
>>>>
>>>> On 03/17/2015 11:25 AM, Dylan Baker wrote:
>>>>> You can either use `piglit resume` to resume the run, or just run
>>>>> any `piglit summary` subcommand to generate one. We could pretty easily
>>>>> add a command to piglit to combine them as well.
>>>>>
>>>>> Dylan
>>>>>
>>>>> On Tue, Mar 17, 2015 at 09:36:01AM -0600, Brian Paul wrote:
>>>>>> Hi Dylan,
>>>>>>
>>>>>> I don't know when this changed, but when piglit runs now, it creates
>>>>>> individual tests/NNNN.json files for each test.  Then at the end, it
>>>>>> combines them into the results.json file.
>>>>>>
>>>>>> If the piglit run is interrupted (host crash/freeze), I'm left with all
>>>>>> those individual files and no results.json file.  Is there command I can
>>>>>> run to generate the results.json file from the individual files?
>>>>>>
>>>>>> -Brian
>>>>>>
>>>>>>
>>>>
>>