[Groff] Fw: pdfmark question

[Keith Marshall]

Nick Stoughton wrote:
> Hi Keith ... I believe you are the author of the pdfmark macros.

Yes, but my preferred address for related correspondence is that at
ntlworld.com, as stipulated in the package, or my SourceForge forwarding
address; using either of these should elicit a quicker response than
writing to me at work.

> I have been using these very successfully recently, but I've come
> across an interesting corner case that I *think* may be a bug, and
> thought I'd run it by you before sending to groff-bugs.

If you write to bug-groff, Werner will most likely simply forward it to
me anyway; this communication will suffice just as well.

> I have a cross reference where the hot-zone word ends up right at the
> right margin. During the first passes, building the href map, the
> pdf*href.mark.end mark appears to cause the word to wrap onto the
> following line. However, during the final "hot-zone" run where the map
> is complete and the last two builds have produced the same document,
> the mark is not added, so the word remains on the first line. However,
> the hot zone is on the following line ...

Sounds like a bug; I can visualise the circumstances which might cause
it, but don't yet have any ideas for solving it.

> I've also got my own script for building the document, based on
> pdfroff.sh in spirit, though started again from scratch. It is not
> impossible that this is what is at fault!

It is unlikely that the driver script would affect the behaviour of the
formatter; I expect that you would see the same, if you did use pdfroff.

> I can send you the pdf to illustrate, but it is very large (3858
> pages)!

No thanks.  A small test case, which I can run through the formatter
myself will be more useful.

> I'm working on trying to produce a small repeatable example, but
> haven't got there yet.
> 
> Any ideas? 

Can't you cut the paragraph in which the fault appears, from your
complete source, and format it stand-alone with the same page layout
settings?  Does that not exhibit the same problem?  If it does, could
you please send it to me, as a test case?

Regards,
Keith.