Automatic coverage testing of mapserv?

[Petter Reinholdtsen]

[Frank Warmerdam]
> I am not interested in changing it without a fairly compelling
> reason.

Well, the reasons I had was to get a system capable of running all the
tests at once and reporting the results at the end (also on the todo
list on wiki), and to reduce the amount of code in the framework.
I've completed a rewrite of the gdal/ and misc/ tests, and need to get
a working python mapscript module to be able to test and convert the
mspython tests.  When this is done, I plan to make a wrapper script to
compile the latest source and run the test suite and finally make some
report on the test results and the current coverage.

When I run the test, slightly more than 20 percent of the code is
exercised, and 11 of 58 tests fail (font problems).  I'm now trying to
add tests to increase the code coverage, but I'm still learning my way
and have a hard time figuring out how I map from a unused
function/feature to a test feature in the map file.  I'll see how far
I can get using the shp2img program, and move to the mapserv program
when I'm unable to increase the coverage using shp2img.

> Hopefully we will have a chance to meet and share a few beers at
> MUM3, and I can do my full "grumpy old man" routine where I berate
> you for wanting to change everything.  Sort of like I did with Sean
> last year.

I would love to share a few beers and watch your routine. :)

Too bad I can't make it to MUM3.  I already have a conference
scheduled in Spain in May, and in Finland in July, so I will have to
tend to my daytime job in June.  :)

Besides, I do not consider traveling to/past USA borders safe these
days, with that current governments lack of respect for human rights
and justice. :/

What about joining us at debconf5?

> I believe that Sean has used the python unittest stuff.

Good.

> I don't know if someone else is picking up on all your patches, but
> I'm not applying them.  I will apply any that seem sensible to me if
> they are submitted properly through bugzilla and end up assigned to
> me.  That is our methodology.

So, what is the mailing list for?  Am I using the wrong mailing list
when I want to get comments on the patches where I am unsure if I
understood the intention of the original code?  I try to create
Bugzilla bugs for the obvious fixes.  I know I could do better here,
and will try to improve.