- From: James Graham <james@hoppipolla.co.uk>
- Date: Tue, 24 Jun 2014 13:46:45 +0100
- To: public-test-infra@w3.org
On 24/06/14 13:43, James Graham wrote:
> On 24/06/14 04:24, Brad Hill wrote:
>> Thanks, Zhiquiang, that was it.
>>
>> It wasn't clear to me that I had to to this step even to run tests that
>> already had a MANIFEST in their subdir - I thought it was just for creating
>> new test suites.
>
> Fundamentally there are two problems here:
>
> There shouldn't be subdirectories with their own manifest files. Is
> there any reason we can't remove these?
>
> The runner should regenerate the manifest automatically if needed. This
> seems quite possible (just make a web frontend for manifest.py that
> checks whether the manifest is missing or the working directory has
> changes compared to the commit for which the manifest was generated and,
> if so, regenerate it).
https://critic.hoppipolla.co.uk/r/1886 for this change, but the code
isn't very well tested ("experimental_include_local_commits" might be a
clue here), so it would be useful for people to try it out before it's
merged.
Received on Tuesday, 24 June 2014 12:47:10 UTC