W3C home > Mailing lists > Public > public-mobileok-checker@w3.org > January 2008

Re: CreateMasterTestOutput options for moki vs testresults.xml

From: Sean Owen <srowen@google.com>
Date: Tue, 15 Jan 2008 15:19:17 -0500
Message-ID: <e920a71c0801151219j66e9b542y2992a2126c857d89@mail.gmail.com>
To: "Dominique Hazael-Massieux" <dom@w3.org>
Cc: public-mobileok-checker <public-mobileok-checker@w3.org>

This was undoubtedly my fault. I typically assume that if I see a
change, and it seems to be causing a test suite to fail, that maybe
the person changing it forgot to update the test. But that's not
always right of course and I should pay more attention to the diff to
see if it makes sense.

I can add this option -- yes this is a good reason. In fact, I think
the default should be to not update testresults.xml, since usually we
will be running this to accommodate changes in moki.xml that are not
relevant to the test's outcome.

On Jan 15, 2008 7:40 AM, Dominique Hazael-Massieux <dom@w3.org> wrote:
> Hi,
> I found at least 3 tests this morning whose testresults.xml had been
> modified in batch after a bug was introduced in the MinimizeTest.xsl,
> and thus gave false results.
> Beyond taking more care when making changes to testresults.xml, I think
> it would help if the CreateMasterTestOutput program could take an option
> to update only either the moki.xml or the testresults.xml.
> (the reasoning being that in most cases, one doesn't want to update
> testresults.xml; moki.xml might need changes more frequently e.g. when
> we add new stuff in the preprocessing).
> What do you think?
> Dom
Received on Tuesday, 15 January 2008 20:19:30 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:21:19 UTC