Home > Archive > PHP Questions and Answers > April 2007 > Re: [PHP-QA] Re: [Gallery-devel] [Fwd: PHP 5.2.2RC1 testing]
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Re: [PHP-QA] Re: [Gallery-devel] [Fwd: PHP 5.2.2RC1 testing]
|
|
| Marcus Boerger 2007-04-11, 6:57 pm |
| Hello Guys,
it might be a good idea to put stuff like those unit tests under the gcov
machine. Maybe it is enough if we have such unit tests run without valgrind.
best regards
marcus
Wednesday, April 11, 2007, 5:53:30 PM, you wrote:
> All 2907 unit tests passed for current Gallery2 svn + mysql with PHP
> 5.2.2-RC-1.
> Thanks,
Best regards,
Marcus
| |
| Nuno Lopes 2007-04-11, 6:58 pm |
| I think this a pretty good idea. We can select a few projects with good unit
tests and run them without valgrind.
I don't have enough time to change the gcov page now, though. Maybe only in
July.. But this really worth it (e.g. I remember when a Zend Framework test
revealed a regression in one of the last RCs)
Nuno
[color=darkred]
> Hello Guys,
>
> it might be a good idea to put stuff like those unit tests under the gcov
> machine. Maybe it is enough if we have such unit tests run without
> valgrind.
>
> best regards
> marcus
>
> Wednesday, April 11, 2007, 5:53:30 PM, you wrote:
>
| |
| Marcus Boerger 2007-04-11, 6:58 pm |
| Hello Sebastian,
is this an area where you could help nuno and me? Meaning setting up
PHPUnit and integrating it in the automatic test runs?
best regards
marcus
Wednesday, April 11, 2007, 11:10:27 PM, you wrote:
> I think this a pretty good idea. We can select a few projects with good unit
> tests and run them without valgrind.
> I don't have enough time to change the gcov page now, though. Maybe only in
> July.. But this really worth it (e.g. I remember when a Zend Framework test
> revealed a regression in one of the last RCs)
> Nuno
[color=darkred]
Best regards,
Marcus
| |
| Antony Dovgal 2007-04-11, 6:58 pm |
| On 04/12/2007 01:57 AM, Derick Rethans wrote:
> On Wed, 11 Apr 2007, Marcus Boerger wrote:
>
>
> I could offfer plenty of code/tests here.
That would be really great.
--
Wbr,
Antony Dovgal
| |
| Antony Dovgal 2007-04-11, 6:58 pm |
| On 04/12/2007 02:12 AM, Derick Rethans wrote:
> On Thu, 12 Apr 2007, Antony Dovgal wrote:
>
>
> We've like 2500 tests for our stuff, but not the machine power to run it
> all under valgrind :) Some of it wouldn't be able to run without some
> database configuration or servers though.
>
> I already experimented with exporting our test results to some database
> as well, but that's still quite primitive (and unfortunately not
> reachable from outside our network).
Well, I'm ready to invest some time in this, if Nuno is busy (and does not mind).
What kind of test are you using?
We probably need to create a separate section of gcov.php.net to see the test results and
include these tests into the run to increase the coverage.
Nuno, what are you thoughts on this?
--
Wbr,
Antony Dovgal
| |
| Sebastian Bergmann 2007-04-12, 3:57 am |
| Marcus Boerger wrote:
> Hello Sebastian,
Hello all,
> is this an area where you could help nuno and me? Meaning setting up
> PHPUnit and integrating it in the automatic test runs?
I would be glad to help with this. I do not have much time until the
end of April, though.
Please note that Sebastian Nohn already has a CruiseControl setup in
place that pulls PHP 5.2 and PHP 6 from CVS, builds it, runs the PHP
test suite as well as Zend Framework, etc. He might also be interested
in helping with this.
> best regards
> marcus
Best regards,
Sebastian
--
Sebastian Bergmann http://sebastian-bergmann.de/
GnuPG Key: 0xB85B5D69 / 27A7 2B14 09E4 98CD 6277 0E5B 6867 C514 B85B 5D69
| |
| Nuno Lopes 2007-04-12, 6:57 pm |
| > Well, I'm ready to invest some time in this, if Nuno is busy (and does not
> mind).
Sure I don't mind. I really appreciate it ;)
> We probably need to create a separate section of gcov.php.net to see the
> test results and include these tests into the run to increase the
> coverage.
> Nuno, what are you thoughts on this?
I'm very bad at designing interfaces.. But a separate section seems fine.
I can give you support on the technical side and review patches, etc.. Maybe
I should also write you a quick guide on how the stuff works (the cron
script, how it posts the data to the web interface,..)
Nuno
| |
| Ilia Alshanetsky 2007-04-12, 9:57 pm |
| Can our existing gcov machine handle that many tests? Having to wait
a w for results would be somewhat inconvenient... Perhaps we need
to upgrade this machine if we plan to extend the test suit?
On 11-Apr-07, at 6:12 PM, Derick Rethans wrote:
> On Thu, 12 Apr 2007, Antony Dovgal wrote:
>
>
> We've like 2500 tests for our stuff, but not the machine power to
> run it
> all under valgrind :) Some of it wouldn't be able to run without some
> database configuration or servers though.
>
> I already experimented with exporting our test results to some
> database
> as well, but that's still quite primitive (and unfortunately not
> reachable from outside our network).
>
> regards,
> Derick
>
> --
> Derick Rethans
> http://derickrethans.nl | http://ez.no | http://xdebug.org
>
> --
> PHP Quality Assurance Mailing List <http://www.php.net/>
> To unsubscribe, visit: http://www.php.net/unsub.php
>
Ilia Alshanetsky
| |
| Antony Dovgal 2007-04-13, 7:57 am |
| On 04/12/2007 09:29 PM, Nuno Lopes wrote:
>
> I'm very bad at designing interfaces.. But a separate section seems fine.
> I can give you support on the technical side and review patches, etc..
> Maybe I should also write you a quick guide on how the stuff works (the cron
> script, how it posts the data to the web interface,..)
Excellent idea =)
--
Wbr,
Antony Dovgal
| |
| Nuno Lopes 2007-04-14, 6:57 pm |
| Although the machine isn't very fast, the upgrade to Etch has provided a
*huge* speedup. For example the PHP 5.2 build time was cut in about 15
hours. If we don't run those tests with valgrind the machine should be able
to run them ok.
Nuno
[color=darkred]
> Can our existing gcov machine handle that many tests? Having to wait a
> w for results would be somewhat inconvenient... Perhaps we need to
> upgrade this machine if we plan to extend the test suit?
>
>
> On 11-Apr-07, at 6:12 PM, Derick Rethans wrote:
>
| |
| Marcus Boerger 2007-04-14, 6:57 pm |
| Hello Nuno,
for the moment i'd say adding tests in non valgrind mode is frin for those
15 hours. That should allow us running tons of tests...
best regards
marcus
Saturday, April 14, 2007, 3:13:53 PM, you wrote:
> Although the machine isn't very fast, the upgrade to Etch has provided a
> *huge* speedup. For example the PHP 5.2 build time was cut in about 15
> hours. If we don't run those tests with valgrind the machine should be able
> to run them ok.
> Nuno
[color=darkred]
Best regards,
Marcus
| |
| Ilia Alshanetsky 2007-04-14, 6:57 pm |
| Running tests without valgrind is a bad idea IMO, even when the test
returns expected results the background code could be corrupting
memory, consequence of which may not need seen inside the test
itself. I'd rather run fewer tests with valgrind, then more tests but
without the full benefit of comprehensive testing.
On 14-Apr-07, at 9:13 AM, Nuno Lopes wrote:
> Although the machine isn't very fast, the upgrade to Etch has
> provided a *huge* speedup. For example the PHP 5.2 build time was
> cut in about 15 hours. If we don't run those tests with valgrind
> the machine should be able to run them ok.
>
> Nuno
>
>
>
> --
> PHP Quality Assurance Mailing List <http://www.php.net/>
> To unsubscribe, visit: http://www.php.net/unsub.php
>
Ilia Alshanetsky
| |
| Marcus Boerger 2007-04-14, 6:57 pm |
| Hello Ilia,
I disagree here. We need to get as close to 100% coverage. But still
that does not mean we get erverything. Running other tests now gives us
the ability to catch much more stuff. And also simply getting 100%
coverage does not mean we get all code paths. And also not necessarily
check for all inputs.... So IMO we should do both.
best regards
marcus
Saturday, April 14, 2007, 4:47:41 PM, you wrote:
> Running tests without valgrind is a bad idea IMO, even when the test
> returns expected results the background code could be corrupting
> memory, consequence of which may not need seen inside the test
> itself. I'd rather run fewer tests with valgrind, then more tests but
> without the full benefit of comprehensive testing.
> On 14-Apr-07, at 9:13 AM, Nuno Lopes wrote:
[color=darkred]
> Ilia Alshanetsky
Best regards,
Marcus
| |
| Ilia Alshanetsky 2007-04-14, 6:57 pm |
| I am not saying we should not aim for 100% code coverage. But, code
coverage is only 1/2 the battle, we need to make sure we are not
corrupting and leaking memory in the course of the execution, which
is information only valgrind can provide to us reliably. That's why
we need it enabled for our test suit as we currently do on gcov, I
don't want to turn it off just so that we could run more tests.
On 14-Apr-07, at 11:48 AM, Marcus Boerger wrote:
> Hello Ilia,
>
> I disagree here. We need to get as close to 100% coverage. But still
> that does not mean we get erverything. Running other tests now
> gives us
> the ability to catch much more stuff. And also simply getting 100%
> coverage does not mean we get all code paths. And also not necessarily
> check for all inputs.... So IMO we should do both.
>
> best regards
> marcus
>
> Saturday, April 14, 2007, 4:47:41 PM, you wrote:
>
>
>
>
>
>
> Best regards,
> Marcus
>
Ilia Alshanetsky
| |
|
|
|
|
|