[Casa-framework] test programs

[Raymond Rusk]

On Fri, 1 Dec 2006, Wes Young wrote:

> I would be just as happy if you put those scripts in something like
>
> xmlcasa/implement/measures/scripts

Ok, that sounds like a good solution.  It also keeps the C++ and Python 
test code separated.  Test scripts that call methods in multiple 
components rather than in just one component interface would be better 
candidates for xmlcasa/scripts/test.

As for unit versus regression testing, the Glish and now Python scripts 
were the only tests that systematically called every method in the 
component's interface so I think they are closer to unit than regression 
tests.  None, as in zero, of the old DO's had C++ test code.  In fact, our 
makedefs didn't support unit testing in the old appglish directory 
hierarchy where the DO's lived, which always irritated me.

If we actually stored old output and compared current runs to past runs, I 
would have to agree that we are regression testing.  But that isn't the 
case, at least for the test scripts that I have been dealing with.  I 
intend to make it so for atmosphere, however, so that I can compare 
TELCAL's output with what we get.

> One could argue and I would not disgree that xmlcasa/scripts/test brings all 
> the components into play so it's not appropriate for single component python 
> test scripts.
>
> My general feeling is a C++ test program is a unit test, where a script is a 
> regression test, YMMV.
>
> So there you go.
>
> wes
> wyoung at aoc.nrao.edu
>
>
> On 1 Dec 2006, at 22:13, Raymond Rusk wrote:
>
>> On Fri, 1 Dec 2006, Wes Young wrote:
>> 
>> > Scripts, I suggest test scripts should go into scripts/test.  So that
>> > spiffy python test script you wrote call tSpiffyTest.py should find
>> > it's way into xmlcasa/scripts/test/tSpiffyTest.py.
>> 
>> Will you be adding a xmlcasa/scripts/test directory then?  In my opinion, 
>> instead of having dozens of scripts in one spot, it would be nicer to have 
>> a test directory where the component lives and put the Python scripts 
>> related to that component in it.  Or course I am biased since that is what 
>> I have done for quanta, measures, coordsys and image.
>> 
>> -- 
>> Dr. Raymond Rusk                                        Dr. Raymond Rusk
>> ALMA Software Engineer                           Ingenieur Logiciel ALMA
>> National Research Council Canada   Conseil national de recherches Canada
>> Herzberg Institute of Astrophysics   L'Institut Herzberg d'Astrophysique
>> DRAO, P.O. Box 248                                    OFRA, B.P. Box 248
>> Penticton, BC V2A 6J9           |\^/|         Penticton, (C.-B.) V2A 6J9
>> Government of Canada         _|\|   |/|_          Gouvernement du Canada
>> Tel: (250)490-4347           >         <              Tel: (250)490-4347
>> Fax: (250)493-7767            >_./|\._<               Fax: (250)493-7767
>> Raymond.Rusk at nrc-cnrc.gc.ca                  Raymond.Rusk at nrc-cnrc.gc.ca
>

-- 
Dr. Raymond Rusk                                        Dr. Raymond Rusk
ALMA Software Engineer                           Ingenieur Logiciel ALMA
National Research Council Canada   Conseil national de recherches Canada
Herzberg Institute of Astrophysics   L'Institut Herzberg d'Astrophysique
DRAO, P.O. Box 248                                    OFRA, B.P. Box 248
Penticton, BC V2A 6J9           |\^/|         Penticton, (C.-B.) V2A 6J9
Government of Canada         _|\|   |/|_          Gouvernement du Canada
Tel: (250)490-4347           >         <              Tel: (250)490-4347
Fax: (250)493-7767            >_./|\._<               Fax: (250)493-7767
Raymond.Rusk at nrc-cnrc.gc.ca                  Raymond.Rusk at nrc-cnrc.gc.ca