I think that we should stick with LaTeX. The current document is in
LaTeX, and the need to use an older version was only to verify that
we had the correct sources. Once we're under source control, that
issue vanishes, and we can easily produce PDF, HTML, and other
formats. As most of the writing is text, the learning curve for
most authors is quite short. And for a standard, cleanly separating
concepts (like emphasis) from ad hoc formating (italics) is quite
important; the more so for large documents. LaTeX is actually
better at that than many other systems (whose WYSIWYG but
undisciplined interfaces encourage ad hoc formatting insertions).
Bill
On Jan 4, 2008, at 2:08 PM, Jeff Squyres wrote:
Per Rich's note about making a consolidated MPI-2.1 document
(including all of MPI-1.2+errata and all MPI-2.1+errata), I think
that's a fantastic idea.
1. What will be the source format for this document (and for
successive MPI documents)? The original MPI documents were written
in the then-current version of LaTeX (Rainer Keller tells me that
you have to downgrade to that version of LaTeX to build the docs).
Are other source formats possible?
LaTeX is fine, but there are only a few people who can write in it;
the learning curve is fairly steep. Other open formats with free
editors are feasible for large documents. For example, OpenOffice
supports templates, natively exports to multiple formats (such as
PDF and HTML), supports digital signatures to guarantee the One
True MPI Standard Document, etc. Others are likely possible, too.
2. PDF is fairly prevalent these days; it would be good if MPI-2.1
(and later) had official PDF's published. I don't have much of an
opinion about official postscript docs anymore; does anyone still
have a requirement for that? Official HTML should also be
published. In making PDF's, it would be quite useful if they were
fully indexed, cross-linked (e.g., clicking in the TOC and index
takes you to the appropriate section), and full-text searchable.
[pdf]LaTeX, OpenOffice, and others can produce such PDF documents.
--
Jeff Squyres
Cisco Systems
William Gropp
Paul and Cynthia Saylor Professor of Computer Science
University of Illinois Urbana-Champaign