[PATCH 1 of 5] run-tests: when building json, use result.failures instead of result.faildata

[Pierre-Yves David]

On 05/08/2015 04:23 PM, Martin von Zweigbergk wrote:
>
>
> On Fri, May 8, 2015 at 11:54 AM Pierre-Yves David
> <pierre-yves.david at ens-lyon.org <mailto:pierre-yves.david at ens-lyon.org>>
> wrote:
>
>     # HG changeset patch
>     # User Pierre-Yves David <pierre-yves.david at fb.com
>     <mailto:pierre-yves.david at fb.com>>
>     # Date 1431066024 25200
>     #      Thu May 07 23:20:24 2015 -0700
>     # Node ID dccc3de8c055f386a1dbfe72e86091251dbd0b50
>     # Parent  c25b2adb3664cd3c488e2c53aab0c64100d40af7
>     run-tests: when building json, use result.failures instead of
>     result.faildata
>
>     It is unclear to me why 'faildata' was used. Lets use the same kind
>     of attribute
>     as for the other groups.
>
>
> I think this needs to be described better. They don't seem to be
> equivalent. For example, "faildata" is not populated with failures that
> were accepted in interactive mode. addFailure(), which adds to
> "failures", is called from two places and I know too little to
> understand what kinds of failures would be added in each place and
> whether the same errors would be added to "faildata"

I've added a test to check for the --interactive case and everything 
seems to be fine with python-27 (I could not get python2.4 to run test 
at all).

I think we are fine here. If we actually fine regression in the wild, 
this will be a good time to add a test case.

-- 
Pierre-Yves David