[Piglit] [PATCH 2/2] results.py: Allow partially completed runs to be summarized

[Dave Airlie]

On 14 November 2014 09:26, Dylan Baker <baker.dylan.c at gmail.com> wrote:
> This restores the ability to create a summary from a partial set of
> runs by using the same functionality as resume.
>
> Signed-off-by: Dylan Baker <dylanx.c.baker at intel.com>
> cc: airlied at gmail.com

Tested-by: Dave Airlie <airlied at redhat.com>

wfm, please push!
> ---
>  framework/results.py | 13 +++++++++++++
>  1 file changed, 13 insertions(+)
>
> diff --git a/framework/results.py b/framework/results.py
> index 175a386..6ab22b1 100644
> --- a/framework/results.py
> +++ b/framework/results.py
> @@ -190,6 +190,19 @@ def load_results(filename):
>      # file descriptors
>      if not os.path.isdir(filename):
>          filepath = filename
> +    elif os.path.exists(os.path.join(filename, 'metadata.json')):
> +        # If the test is still running we need to use the resume code, since
> +        # there will not be a results.json file.
> +        # We want to return here since the results are known current (there's
> +        # an assert in TestrunResult.load), and there is no filepath
> +        # to pass to update_results
> +        # XXX: This needs to be run before searching for a results.json file so
> +        #      that if the new run is overwriting an old one we load the
> +        #      partial and not the original. It might be better to just delete
> +        #      the contents of the folder if there is anything in it.
> +        # XXX: What happens if the tests folder gets deleted in the middle of
> +        #      this?
> +        return TestrunResult.resume(filename)
>      else:
>          # If there are both old and new results in a directory pick the new
>          # ones first
> --
> 2.1.3
>