Hello, On Mon, 17 Oct 2011 10:58:55 -0400 you wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 10/17/2011 10:21 AM, Adam GROSZER wrote:
Hello,
On Mon, 17 Oct 2011 10:08:55 -0400 you wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
We've lost our discipline about keeping the daily tests passing. I think this is due to two factors:
- - We are running tests for code that nobody is actively maintaining (e.g., all the 'z3c.*' packages which have been broken for months).
If you mean the failures on windowze, I think most of them fail on lxml not having windows egg :-S, someone should kick Sidnei(???)
If the z3c.* packages are being maintained, but not by folks who can resolve that issue on Windows, then we should do the testing on a platform where the maintainers *can* get the tests to pass. Having the board stay red destroys its value for *everybody*, including those who don't need or care about those packages.
- - Because of the noise of those failures, we've gotten complacent about seeing multiple failures in the daily test report, and thus quit trying to find and fix "new" failures in code which is maintained.
I would like to propose that we quit reporting test failures for the unmaintained packages ('zc3.*', for a start. If somebody really thinks there is value in such testing, we could perhaps run those tests only weekly (like the others which have "expected" failures).
If it's about suppressing the *reporting* then suppress that, not the test runs please. Just in case someone wants to maintain a package ;-)
Why waste any resources on tests if nobody is going to find and fix failures?
Aber hallo, z3c.form is in there. I do *not* think that's not maintained. Actually I *know* what's the problem, just can't solve it. Keeping the tests running is like having the *positive* attitude instead of the dark "I don't care". -- Best regards, Adam GROSZER -- Quote of the day: Drama often obscures the real issues