Re: Knowing which tests are in the repository

On Wed, Aug 21, 2013 at 3:15 PM, James Graham <james@hoppipolla.co.uk> wrote:
> Yes, that does work, but in practice your tests have been problematic on
> automated infrastructure due to the long run time; the DOM Range tests are
> an example of this, I believe. It would be preferable to be able to specify
> that they should run in chunks without either forcing you to break them into
> multiple files, or to reduce coverage, or to have additional
> project-specific annotations chunking the tests.

I'm not sure the long runtime *per file* has caused big problems with
my tests.  High memory usage for some of them, and high overall
runtime for the suite, have been the real issues I'm aware of.

On Wed, Aug 21, 2013 at 8:10 PM, James Graham <james@hoppipolla.co.uk> wrote:
> If we are not already, we should encourage reftest authors to reuse
> references wherever possible rather than creating new files.

I can attest that this also often makes the test easier to understand
and review.  In my experience with transforms, an awful lot can be
easily written to test against a standard 100px-by-100px lime square,
and in such cases you don't even have to look at the ref to understand
the test.  Even when not, it still makes it a lot easier to review a
test suite if a series of related tests uses one ref.

Received on Friday, 23 August 2013 11:00:12 UTC