Handling of ancestor *info content in manpages output

["Michael(tm) Smith" <smith@sideshowbarker.net>]

Joe Orton <jorton@redhat.com> writes:

> w.r.t. the AUTHOR stuff, it doesn't seem to pick up the author element 
> from /book/bookinfo/author, which is the only place in my manual I have 
> such a beast.

I just checked in a change that, for each Refentry, picks up data
for the AUTHORS section from the *info children of all ancestors of
that Refentry (along with the *info child of the Refentry itself).

Along with checking for author elements, it also checks for all
the following: editor, collab, corpauthor, corpcredit,
othercredit, orgname, publishername, publisher.

I guess in some cases this might result in users ending up with a
lot more stuff in the AUTHORS section than they might want. For
one thing, if you have the same author/editor/whatever info in
your source at multiple places in the hierarchy (for example,
specified in bookino, referenceinfo, refentryinfo), it will appear
multiple times in the section.

Also, if you don't have any context information for an author or
other contributor explicitly specified within a personblurb or
authorblurb or contrib element, it may not be clear in your output
who authored which part. For example, if you have authors
specified at multiple levels in a book, and you don't have any
*blurb or contrib description marked up for any of them, you would
get this by default:

  Benvenuto Cellini << [from Bookinfo]
    Author.

  Joe Orton         << [from Referenceinfo]
    Author.

  Michael Smith     << [from Refentryinfo]
    Author.

Which is going to look more and more odd the more authors you
have. But if you mark up *blurb or contrib descriptions for each
of those you end up with:

  Benvenuto Cellini
    Principal author for the overall guide to Foo.

  Joe Orton
    Main author and editor for the portion of the C API
    documentation for Foo.

  Michael Smith
    Author for the specific document you are now reading.

smime.p7s