Here is a proposal for a script testing MSL/conjure.
It compares an image created with MSL to another image created with
If you see any red pixels in the result the test failed.
It's a quick hack but it might be all that is needed.
On Mon, 29 Sep 2008, Bob Friesenhahn wrote:
> On Mon, 29 Sep 2008, Max wrote:
>> You mean for regression testing?
>> I'm just wondering how such a test looks like for an image processing
>> You would have to compare the resulting image with an image that is
>> known to be correct, right?
>> Can you give me an easy example?
> I am not sure how to compare images using MSL. The PerlMagick test
> suite uses two means of comparison. In some cases the computed image
> signature string is used. In other cases the output image is compared
> against a reference image and some small error is allowed.
> For the utility tests, there is no comparison at all. The test passes
> if it runs through to completion without reporting an error. Visual
> inspection of the results is necessary.
> Since we already have the gruelling PerlMagick tests, the MSL tests
> could be of the type where someone has to inspect the result to see that
> they are correct. In this situation, the MSL "tests" also act as a demo
> and an example. The person doing the inspection needs to know what the
> results should be.
>> On Mon, 22 Sep 2008, Bob Friesenhahn wrote:
>>> A way that you can help improve conjure quite a lot is to contribute
>>> some formal tests for it. At the moment there are no MSL test scripts
>>> included at all, which means that MSL is not getting tested.