#26 Failed Builds always rebuilding at next polling period

[Michael Ferrante]

From draconet-developers list:

Excerpt
-------
If draco triggers a build which subsequently fails I'm now
seeing the same build being kicked off again at the end
of the next polling period even if no check-ins have
occurred in the interim. The build obviously fails again
(since nothing has changed) so it goes into an endless
build/fail loop. The only way out of the loop is to correct
the code so that the build is successful, at which point
polling detects no further changes as expected.

Full Message
------------
I'm currently trying out the new 1.6 beta in combination
with VSS for source control and Nant for doing the
builds. In addition to the possible problem with the
ignore comment feature that I reported yesterday, I've
just spotted something potentially far more problematic:

If draco triggers a build which subsequently fails I'm now
seeing the same build being kicked off again at the end
of the next polling period even if no check-ins have
occurred in the interim. The build obviously fails again
(since nothing has changed) so it goes into an endless
build/fail loop. The only way out of the loop is to correct
the code so that the build is successful, at which point
polling detects no further changes as expected.

What's more, the failed build output from Nant now
appears to be being written to the draco.log, even when
the trace level switch is set to 1, which is causing the
log to grow at even faster rate than normal (as well as
making it more cluttered).

Any thoughts?

Phil

[Nobody/Anonymous]

Logged In: NO

I am also seeing this behavioun with VSS.

mfletche@petro-canada.ca

[Nobody/Anonymous]

Logged In: NO

Beta-2 seems to fix the problem.

mfletche@petro-canada.ca