Re: [lmi] Conclusions drawn from certain 2007 xsl-fo experiments

[Greg Chicares]

On 2008-11-16 23:45Z, Greg Chicares wrote:
> We need to automate regression testing for pdf output.
[...]
> our best option for regression testing is apparently to
> compare xsl-fo files.

As of 20081212T0200Z, we can do this:

cd /opt/lmi/bin
./lmi_cli_shared \
  --accept \
  --cnsfile=/path/to/whatever.cns \
  --data_path=/opt/lmi/data \
  --emit=emit_pdf_file \
  --pyx=xml.system_testing \
  2>&1 |less

for instance to test the 20081212T0201Z change.

That ought to help us to start thinking more concretely
about how to extend system testing to encompass output.

It might also help us to track down an intermittent
error that a couple of users have reported recently: in
some already-crowded sections of a pdf file, each column
of numeric data appears several times, with copies shifted
laterally. Digits overlay each other. To me, it looks like
a nixie tube
  http://www.tube-tester.com/sites/nixie/data/in-18/in-18-f.jpg
with power applied to more than one cathode. Anyway, the
first step is probably to determine where in this pipeline:
  C++ --> xml --> xsl-fo --> pdf
the error arises; that's where the command above may help.