Re: [boost] Beta posted for new regression tests

From: David Abrahams (dave_at_[hidden])
Date: 2002-07-24 09:03:53

From: "Beman Dawes" <bdawes_at_[hidden]>

> At 08:49 PM 7/23/2002, David Abrahams wrote:
>
>  >> Is it possible to add status/Jamfile entries for at least a few of
the
>  >more
>  >> important tests?  That would be a start.
>  >
>  >Did you read what I asked? I can add entries, but they won't use the
same
>  >rules so they won't leave the same residue.
>
> I don't understand that; it seems pretty surprising that an entry like
> "run" would have different behavior because the target is in the python
> library.

You're still missing my point. The *rules* that run Python tests are not
currently the same ones that are used to run the status/Jamfile tests. In
other words, for tests which invoke Python I'm not using the "run" rule,
even though the test does run an executable. That could potentially be
changed, but it'll take some work.

> But if you say so, we won't use the regular status/Jamfile.

Which Jamfile it's in is irrelevant. What matters is which rules are being
used to set up targets, dependencies, and actions.

>  > I'll ask again: what residue
>  >are you using?
>
> In the target directory, the .output, .success, .failure, and .test
files.

Oh :(
I guess I have no choice but to integrate the Python tests with testing.jam
then. It's probably just as well, but I wasn't ready for this right now...

>  From the bjam output, the compile, link, and run messages.
>
>  From the Jamfile, the "run", "run-fail", "compile", "compile-fail",
> "link", and "link-fail" entries.

Really, you're scanning the Jamfile?

:( :(

That makes modularizing very difficult. If you want the testing rules to
generate a file that describes all the tests which get run, that would be
relatively easy to do, I think...

-Dave