Re: [pdf-devel] filter testcases

[jemarch]

   Using the pdf_filter seems the best option to me. I didn't know about
   test drivers, but I'm reading about Dejagnu and seems usefull. However,
   I think that probably we can drive the test cases from a simple
   makefile, as using pdf_filter would nees just to filter some data and
   diff it with an already checked correct output.

I agree in that since pdf_filter is a batch-oriented tool Dejagnu
(expect) could be more than we really need.

   Maybe the worst part of this is just design the concrete tests cases
   -to ensure that they control every single limit and error prone case-,
   which is a bit boring, but I think that by using the pdf_filter it must
   be quite straightforward to build them.

My perception of the issue is that it should be easy to write some
generic mechanism (maybe a shell script) to test filters using
pdf_filter: just encode something, decode it and then compare with the
original data. Then we could call that script in the `make check'.

I am thinking in something like this:


-- test-filters.sh

# Generic stuff
...
test_filter() {

  ...

}

# Perform the actual testcases calling pdf_filter
test_filter(ascii85, input_test_file_1)
test_filter(asciihex, input_test_file_2)
...

# Return final status to the testdriver (Makefile)

-- end of test-filters.sh

What do you think?