[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: (Partial) implementation of new math encodings

Matthias Clasen writes:

 > I wonder how one could check that everything is definitively ok - apart

there is nearly no way to check that everything is okay :-)

 > from just having a close look at the dvi output. Are there any clever
 > methods for finding sublte differences between two dvi files ? Computing
 > the difference of the bitmaps ?

for latex2e we do use a huge set of regression test files that do
produce certified output that can automatically be compared (comparing
log files). unfortunately that scheme does not work in this case as
something like


will produce completely different results since the logical glyph
positions are recorded.

but one idea of my head:

as long as the implementation does used virtual fonts, what about this

for F in testfiles ; do

 - produce F1.dvi from F using current encoding scheme
 - produce F2.dvi from F using new encoding scheme
 - run F2.dvi through the program that removes the virtual fonts
   from the dvi (was it dvi2dvi?) resulting in F3.dvi
 - compare F1.dvi with F3.dvi perhaps by first converting the file
   via dvitype or something to a more human readable form
   (or perhaps try cmp first and if the files differ do diff of 
    ascii forms of the dvi files)

just some thought ...