Re: Errors in the test suite

Hello !

Trying to see what were the capabilities of a MathML rasterizer named JEuclid,
written in Java, i ran it through the test-suite. I needed to add the DTD
because Jeuclid tries to validate each file before conversion. I know there is a
catalog that can be used, but i can't seem to get it to work, but this problem
is only mine ;-)

For those who may be interested, the results are available on the web and sum up
what works and what does not in Jeuclid 2.0. Another thing i noticed during the
tests, were the failures/missing things that i think were independant of Jeuclid
: they are to be found in the 'bad' and 'alone' section. It can be reached at
the following address :

The 'alone' section lists mml files without a reference image, which makes it
impossible to compare with the rendered image. I think that the 'bad' sections
regroups the stuff "should" NOT work, but i put some files in tht may have some
mistakes too (and maybe shouldn't, ie non bad*.mml files). These are the
following : emptyContent1,  mactionBhigh1, mathAdisplay1, mathAdisplay2,
mathAmode1, mfracZComp1, nestAction1, nestedMath3 and maybe noChildContent and
tooFewContentContainer. You can see the log and error output in the bad/failed
section and alone/failed section.

My aim here is not to criticize the work done in this testsuite, which is really
great. I just want to spot some things i find bizarre, and i would like to know
if the failed files are the responsability of the rasterizer or some error in
the test-suite. What i trying to get is a comparison of some MathML-aware tools,
to have a reliable basis on which one can choose a tool in function of the
quality of the output we want.

Thank you all for the good job, i hope i do not "disturb" too much ;-)
Have a nice day,

Received on Tuesday, 27 April 2004 21:24:08 UTC