>
> 2) I'm not sure about the pipeline. It does seem like a potential
> candidate, but I'm not sure what conditions control use of the pipeline. I
> know that I get good output interactively regardless of whether I turn on
> profiling or not. Is there a way to know for sure, on any particular
> publish, whether the pipeline is active?
I'm not sure how to *detect* it, but I know that profiling and DLM trigger
the pipeline. I'm thinking there is a third, common trigger, but it's
escaping me now. One of the PDFs (or Help Center) might have more on what
functionality triggers it.
> 3) I'm not sure about the parsers, but my recollection is that one is
> primarily for SGML docs and the other is for XML docs. Plus, I doubt it has
> anything to do with processing instructions in the doc--my tests are based
> on the same document, with no changes between tests, so the same PI's would
> be there under all conditions. (The doc is in source control, so Arbortext
> can't change it even if it wants to.)
It was/is a reach, but I was wondering if one parser is used always in batch
while in interactive, a choice may be made depending on various things, one
of which is whether or not Arbortext had read it before. I was thinking if
you could force the parser to flip from one to the other in one of the
modes, you might see different behavior. Admittedly this idea was also
paired with recollections of problems we had with some character entities,
charent.cf, and the pipeline prematurely resolving character to font and
losing that before the end of the pipeline.
Are the apostrophes rendering in the PDF in the same typeface as each other
and the rest of the text?
--
Paul Nagai
