On Wed, 2002-06-12 at 22:50, Brekke, Jeff wrote:
> Martin,
> 
> If someone chose to implement something like this, the testing stuff
> wouldn't change at all.  Create another jar target only that doesn't execute
> the unit tests.  

Here we are communicating in different languages ;). I just need the
tests, even though I know they fail, so I need them to execute...
I also don't mind calling another target, as long as the tests can fail
and my build doesn't. But that still means duplicating the test
build.xml, since it cannot but fail (hmm is that correct english..).

So maybe we can do with an extra target that calls do-test-dontfail (or
something similar), when using the target maven:testdonfail from the
main build.xml (or is this what you meant in the first place?). If I
have a +0 or higher from you on this, I will make a patch for it..

And don't worry failing will be the default ;) (that is what I want when
developing, but my automated builds don't like failing too much, since
the site will never get an update that way, which was the reason in the
first place to make that simple hack...)

This would done be in core/build.xml and wouldn't be too
> much duplication.
> 
> I really hope that having the tests fail the build when the tests fail
> remains the default behavior of Maven.  

Yes please.. I wouldn't want to change the default behaviour in any way.
> One of Maven's aspects that attracted me was it was/is a collection of best
> practices with regards to build management/development collected from
> various developers experienced in open and closed source java projects.  I
> hope Maven doesn't turn into ant or a super-duper build tool with the
> flexibility to build anything that ever needs building.  This is probably
> why I keep -1'ing this small change wrt failing tests.  

The problem with this best practice is, is that it kills all the
feedback you can get via a nice site that something is wrong and what is
wrong. (which is, at least for me, one of the reasons why I use maven in
the first place). I want my development site to reflect the latest
nigthly build and not the latest without failing tests. We are still
dealing with people here, that somethimes just don't run tests before
they commit. There are enough projects (also here on jakarta) that have
failing tests after a commit, just because someone forgot to run the
tests or forgot to commit something. 
Getting a nag mail from gump is also nice, but is highly outclassed by
the feedback maven can give us if it can actually finish a build where
tests are failing.

Enough on this subject ;) Just give me your vote and I will make a patch
depending on that vote ;)

Mvgr,
Martin



--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to