Re: Autotest, proof-of-concept [was: About TDD and Pharo]

Laurent,

I have similar needs. I would like to extend SUnit before and after each test method. I was wondering about this:

- Introduce a registration mechanism in SUnit (a bit like OB commands)
- TestResult>>runCase: aTestCase can be enhanced by executing what has been registered before and after
- A class TestCommand contains a method #before and #after. TestCommand define inst var #currentTestCase #currentTestMethod

Advantages:
- minor addition to SUnit


We could then refactor the SUnit history to use a command.
It would then be easy to have a logger or something.

Does it make sense?

cheers,
Alexandre

On 29 Jul 2010, at 11:08, laurent laffont wrote:

> Hi,
>
> Autotest use a wrapper to count hits on changed method. Take a look at Autotest>>#findRunAndShowTestsOf
>
> and replace:
>
>       counter := AutotestHitCounter on: changedMethod.  <- this wraps the method
>       [aTestResult := runner run: testMethods]
>        ensure: [counter uninstall].
>    aTestResult hitCount: counter hitCount.
>
> by
>
> aTestResult := runner run: testMethods
>
>
> to check that the problem is here.
>
>
> I haven't tried Cog yet. Is it working out of the box ?
>
>
>
> Laurent
>
>
> On Thu, Jul 29, 2010 at 10:48 AM, Tudor Girba <[hidden email]> wrote:
> Ah, indeed! I forgot ... sorry for the noise :)
>
> Cheers,
> Doru
>
>
>
> On 29 Jul 2010, at 10:45, Henrik Johansen wrote:
>
> On Jul 29, 2010, at 10:40 55AM, Tudor Girba wrote:
>
> Hi,
>
> It looks like Autotest is crashing on a Cog VM/image. Any ideas as to why that happen?
>
> Cheers,
> Doru
>
>
> If he went for using MethodWrappers as described below, it will crash on Cog, as it does not support objects as methods yet.
>
> Cheers,
> Henry
>
>
>
> On 27 Jul 2010, at 10:11, Alexandre Bergel wrote:
>
> Indeed I also want to log for each test:
> - min / max / mean execution time
> - time to first failure
> - % of errors/failures/success
> - run count
>
> That would be cool
>
> so with these datas we know long tests. I think about wrapping run test methods with an object which then can collect these datas. Would you go this way ?
> (Another way is to modify TestResult / TestCase, but it's more intrusive).
>
> Over the last few month I intensively used method wrapper (a.k.a object as compiled method). Time to time, the image just freezes or crashes. Maybe due to the garbage collector. Modifying SUnit should not be that complex. It would be nice to turn SUnit into something more extensible. One shoot two targets.
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
> --
> www.tudorgirba.com
>
> "One cannot do more than one can do."
>
>
>
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.






_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

> Is Annoucements framework is synchronous or asynchronous ?  If synchronous, using Annoucements isn't less intrusive and more extensible ?

:-)
synchronous.
http://code.google.com/p/pharo/issues/detail?id=2721
SLICE-Issue-2721-AnnouncementsForSUnit-Alexandre_Bergel.1 in PharoInbox

I quickly wrote this. I haven't used it yet.

Cheers,
Alexandre


>
> On 29 Jul 2010, at 11:08, laurent laffont wrote:
>
> > Hi,
> >
> > Autotest use a wrapper to count hits on changed method. Take a look at Autotest>>#findRunAndShowTestsOf
> >
> > and replace:
> >
> >       counter := AutotestHitCounter on: changedMethod.  <- this wraps the method
> >       [aTestResult := runner run: testMethods]
> >        ensure: [counter uninstall].
> >    aTestResult hitCount: counter hitCount.
> >
> > by
> >
> > aTestResult := runner run: testMethods
> >
> >
> > to check that the problem is here.
> >
> >
> > I haven't tried Cog yet. Is it working out of the box ?
> >
> >
> >
> > Laurent
> >
> >
> > On Thu, Jul 29, 2010 at 10:48 AM, Tudor Girba <[hidden email]> wrote:
> > Ah, indeed! I forgot ... sorry for the noise :)
> >
> > Cheers,
> > Doru
> >
> >
> >
> > On 29 Jul 2010, at 10:45, Henrik Johansen wrote:
> >
> > On Jul 29, 2010, at 10:40 55AM, Tudor Girba wrote:
> >
> > Hi,
> >
> > It looks like Autotest is crashing on a Cog VM/image. Any ideas as to why that happen?
> >
> > Cheers,
> > Doru
> >
> >
> > If he went for using MethodWrappers as described below, it will crash on Cog, as it does not support objects as methods yet.
> >
> > Cheers,
> > Henry
> >
> >
> >
> > On 27 Jul 2010, at 10:11, Alexandre Bergel wrote:
> >
> > Indeed I also want to log for each test:
> > - min / max / mean execution time
> > - time to first failure
> > - % of errors/failures/success
> > - run count
> >
> > That would be cool
> >
> > so with these datas we know long tests. I think about wrapping run test methods with an object which then can collect these datas. Would you go this way ?
> > (Another way is to modify TestResult / TestCase, but it's more intrusive).
> >
> > Over the last few month I intensively used method wrapper (a.k.a object as compiled method). Time to time, the image just freezes or crashes. Maybe due to the garbage collector. Modifying SUnit should not be that complex. It would be nice to turn SUnit into something more extensible. One shoot two targets.
> >
> >
> > _______________________________________________
> > Pharo-project mailing list
> > [hidden email]
> > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
> >
> > --
> > www.tudorgirba.com
> >
> > "One cannot do more than one can do."
> >
> >
> >
> >
> >
> > _______________________________________________
> > Pharo-project mailing list
> > [hidden email]
> > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
> >
> > _______________________________________________
> > Pharo-project mailing list
> > [hidden email]
> > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
> --
> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
> Alexandre Bergel  http://www.bergel.eu
> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>
>
>
>
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.






_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

 On 29.07.2010 17:45, Alexandre Bergel wrote:

Wow ! Fast !.

That means it should be easy now to do a lot of exciting stuff. Show tests progression, errors/failures as soon as they occur, log several things, drink cocktails,  meet beautiful girls ....that's really cool.

I will be in winter time in 1 week time. We have to enjoy now

Alexandre

Nice!
Maybe it's just me, or could we not add Announcement to the end of every Announcement subclass name though?
Like with the window announcement, I like them better as verbs describing the event, f.ex.
someAnnouncer on: TestStarted do: []
someAnnouncer on: TestEnded send: #x to: y

Cheers,
Henry

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

> Maybe it's just me, or could we not add Announcement to the end of every Announcement subclass name though?
> Like with the window announcement, I like them better as verbs describing the event, f.ex.
> someAnnouncer on: TestStarted do: []
> someAnnouncer on: TestEnded send: #x to: y

Someone else vote for this?

Alexandre
--

_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.






_______________________________________________

Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

> TestResult when: TestCaseStarts do: [:event | do some stuff ].
>
> TestResult when: TestCaseEnds do: [:event | do some stuff ].

Currently, the announcer is stored in TestCase. For me, it looks more intuitive to do
MyTestCase announcer on: TestCaseAnnouncement do: [:ann | ... ].

than dealing with TestResult.

I am okay to move the announcer in TestResult, but that's look a bit to me. Don't you feel?

Cheers,

Alexandre
--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.






_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project