Re: [Groff] pdfmom grep (was parallel text processing)

[Ralph Corderoy]

Hi Peter,

I think you're smuggling a -k or -K through to the first groff that
pdfmom runs.  Here's its -Tpdf pipeline again.

    groff -Tpdf -dPDF.EXPORT=1 -mom -z $cmdstring 2>&1 |
    grep '^.ds' |
    groff -Tpdf -mom - $preconv $cmdstring

The problem is grep seeing invalid UTF-8 and thus deciding stdin is
binary.  A preconv(1) would turn your UTF-8 troff source into
ISO-8859-1, and any non-ASCII characters in that would probably be
invalid UTF-8.  But pdfmom has tried to spot -k or -K in its arguments
and arrange for them to be moved from $cmdstring to $preconv and so used
only by the second groff.  If it's simplistic argv[] parsing has failed,
because you've -xyzk for example, then your -k remains in $cmdstring and
affects the first groff.

Do that strace command again and look at what's being run for that
pipeline.  That should show the problem.

> Argh.  Why don't they make special glasses that let
> you see code as if for the first time whenever you put them on?

Like those for proofreading?  :-)

-- 
Cheers, Ralph.
https://plus.google.com/+RalphCorderoy