Defacto tests (Was: Tentative tests)

On Wed, May 24, 2017 at 7:05 PM Anne van Kesteren <annevk@annevk.nl> wrote:

> On Wed, May 24, 2017 at 5:54 PM, Rick Byers <rbyers@chromium.org> wrote:
> > In terms of the semantics of directories, haven't we said previously
> that a
> > "defacto" sub-directory (or filename suffix) is OK for tests which pass
> in
> > practice but aren't properly specified yet?
>
> I haven't seen that in the documentation.
>
>
> My suggestion to Mike West is put tentative in the documentation (and
> maybe some of the test runner stuff needs updating to ignore these
> resources although I guess that thing was broken anyway). That makes
> it more final and provides a clear path for any appeals.
>

It looks like "defacto" hasn't been discussed on this list. A top-level
defacto/ directory would be a bit worrisome if nobody wanted to maintain it
and try to explain with specs over time. Inside well-maintained test suites
for well-maintained specs I think it'd probably be fine, though,
functioning like a TODO and a regression test at once.

I've not been involved in any of the previous discussion, are there a few
concrete examples by which to judge this idea?