At David Lauzon's suggestion, I just re-ran the unit tests and they were fine. He believes there's a few places in the test code he
can better capture and log "network failures" ... but otherwise, there's no sign of any trouble with the code itself.
Just for consistency with the original "ready to test" note, we will still count 1 Dec 2005 -- 03:37 (UTC) as the "official" release candidate.
It is identical to the 06:46 except the later summarizes 3 additional unit tests suites.
Jeffrey, I'm not sure how you run the API Scan's ... but you can run on both builds if that's ok. (that is, 1 Dec 2005 -- 03:37 (UTC) and the 06:46 one).
They should be identical, except the later publishes more JUnit summaries with the fixed master test script.
(The master test script, that changed in the later one, is not really part of the delivered build package, but is
version v20051201_1315 of releng.wtpbuilder. I suspect you, Jeffrey, already have the right tests scripts on your
scanner machine, from wtpbuilder head?)
If this proves confusing, and there's no difference to AP scanner code, I can remove the 06:46 build?
Thanks all.
| David M Williams/Raleigh/[EMAIL PROTECTED]
Sent by: [EMAIL PROTECTED] 12/01/2005 04:04 AM
|
|
Shortly after I sent my last note, I noticed there were three of our unit tests suites that were not running in the builds.
I tracked down to some test scripts in our wtp-builder that had never been "released" to the wtp-build process.
See https://bugs.eclipse.org/bugs/show_bug.cgi?id=118766
for details.
So, I released the fixed scripts, re-ran the build, our 3 missing unit test suites did run, and ran fine, but there was a failure in one of the WSI Unit tests.
I am not sure why it failed, but am certain its not related to the changed test scripts (well, as I am ever certain about anything ant related).
I will leave to WSI team to investigate and report on that failure.
The actual * code * in the two builds is identical, as evidenced by the
identical map files (on build pages as directory.txt file).
So, both builds should be identical, and I'll assume the test failure is some
intermittant failure?
But, if you have to decide and pick one, please use Thu, 1 Dec 2005 -- 03:37 (UTC) as the "official" RC1 candidate.
If we promote a build without further fixes/builds, it will be that one.
I suspect Jeffrey can "coordinate" the API reports so they make sense (and include the three
missing unit tests that were not running as part of the build).
Thanks all, sorry for the churn.
_______________________________________________
wtp-dev mailing list
[email protected]
https://dev.eclipse.org/mailman/listinfo/wtp-dev
_______________________________________________ wtp-dev mailing list [email protected] https://dev.eclipse.org/mailman/listinfo/wtp-dev
