In using poppler within cairo's test suite to test cairo's PDF
backend, we sometimes run into bugs/limitations in poppler that
introduce false negatives into the test results. That is cairo's test
suite detects a failure, even though the generated PDF is correct.

I want to be careful to always open poppler bug reports for such
issues, so that we don't lose them. Often, in the past, the poppler
community has been extremely responsive to bugs found this way, and
we've been able to get them fixed almost immediately, (thanks, Jeff!),
and simply ensure that we use bleeding-edge versions of poppler when
testing cairo. I trust that the poppler maintainers will continue to
be as responsive, but as cairo's PDF output starts to use more of the
functionality of PDF, it might be harder to keep the pace.

So, my plan is as follows. Whenever we identify a poppler bug exposed
by the cairo test suite we will do the following:

1. Verify that cairo's PDF output is valid by manually running an
   alternate PDF viewer.

2. Report a bug against poppler here with a PDF file, and good and bad
   output images.

3. Annotate this bug to depend on the new bug.

4. Disable that specific PDF test in cairo's test suite.

5. Allow the cairo development to continue.

Then once the bug is fixed, we can re-enable the test in cairo's test
suite. And it will be important to close that loop as quickly as
possible since we'll be leaving a window open for regressions to sneak
into cairo.

Also, as we add cairo features that trigger bugs like this, it will be
important to get poppler fixed up so that people can use tools like
evince to view their cairo-generated PDF files correctly.

Once again, we love poppler and really appreciate all the work you do
on it!

Other than the dependency notices here, I don't expect much activity
on this bug report. It's really just a placeholder.