Hi,
I think Jogi and you discussed different things.
To check the full functionality of a build you have to run all automated
test scripts. This is done by the Sun Automation team. The team run all
test scripts in 3 levels and on 3 builds (e.g. Level 1 on build m197, 2
on m198 and 3 on m199). So after 3 builds you will have the results of
all test scripts and you can get an overview about the general status
of the version.
At a release (OOx) the number of test scripts are reduced. This makes it
possible to run the test in a time frame of one day, because the builds
after code freeze or RC builds come sometimes very often. But the test
cycle with all test scripts are still running on the Master-Builds
(SRC).
At last builds only sanity checks should be run.
On Master we run all tests on English only. The release tests are done
for all languages.
To check the full functionality of a new language and/or a new platform
it is needed to run all tests at once. Only with all tests you can
identify possible problems in the builds.
I know, that all tests are not running automatically on new languages
or new platforms. But the Sun team is working on documentation that the
L10N teams or other community members can make the changes by
him-/herself.
I also know, that it isn't possible in most L10N teams to run all tests
on all platforms. So it often help, that general functionality is
language independent. But all tests with language dependent
functionality has to be run. Therefore the information about these
test scripts will be collected by the Sun team.
To get a full overview about all states of all languages and platforms
a central point of contact should be created inside the QA-project. I
know, on the last OOo conference Andre and Jogi discussed this. But as
I can see here, this tooling will be more and more important for the
team.
Thorsten
André Schnabel wrote:
Hi Jörg,
Joerg Sievers schrieb:
Hi Andre,
Andre Schnabel wrote:
I used to run the *_upd.bas tests per module. If there are more tests
than that, please tell (maybe at next weeks IRC meeting ;-) ).
in each modul _all_ tests need to be run. All *.bas files.
If there is an old, unsupported test it should be removed in CVS
branch by maintainer of the test (see description in .bas) file.
I thought, the answer is something like that. Excuse me, but thinking of
the recent discussions and knowing that many of the tests will fail in
localized environments.
All the discussions about localized testing make me wonder. At the one
hand we are told, that functionality doesn't need to be tested in a
localized environment. SBA stressed the question "what *not* to test" in
the last IRC meeting.
At the other hand I now hear that we would need to run all the tests.
Imho both of this statements cannot be true at the same time.
I hope, we can sort this out.
André
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
