On Tue, Feb 05, 2008 at 09:58:39AM +0100, Thomas Klausner wrote:
>
> Therefore, my current approach would be to hack together a few scripts
> that fetch stuff from git, run the tests, collect the results, and
> report them.
That's what I did at $work. I store all the results in a database, and
hope to come up with creative reports (especially historical ones) in
the future. TAP::Parser was very helpful in collecting the test results.
My smoke script is basically fetching the list of all test files,
then looping over each of them with TAP::Parser, and producing
various objects (stored in a db) for smoke / script / test.
The biggest trouble I had was for diagnostics. I ended up considering
that diagnostics output after a test result belong to the test result
(as a comment to it), and that diagnostics appearing before the first
test result are "global" to the whole test script. Which means that "#
Looks like you failed 3 tests of 22." is always attached to the last test.
--
Philippe Bruhat (BooK)
You are never too old to have your seat tanned by your Grandmother.
(Moral from Groo The Wanderer #41 (Epic))
