[pracjourn-forum] Comments on Art Ogawa's paper

[Will Robertson]

Hello

Art's paper for this issue of TPJ raises some very interesting and well 
thought out questions, some of which I'd like to comment on here to 
possibly raise some discussion. Since Art is running this list and it's 
his paper, I don't have a problem sending this out before the weekend 
even though some people may not have subscribed yet to the list!

First of all, and I may need to be corrected on this, I believe MikTeX 
can automatically detect when a package is requested in a document run 
and download it without user intervention -- or even user knowledge. 
This functionality, however, has been provided by a complete, and 
manual, repackaging of the relevant parts of CTAN. Hopefully the new 
changes to be effected to CTAN will provide a more automatic method for 
this, and allow this functionality to be ported to the other major 
operating systems.

Secondly, on fonts and operating systems. The Latin Modern fonts are 
now also available in OpenType format, and so if indeed they become 
LaTeX's default fonts (as is already the case with ConTeXt) then these 
may be installed (well, a subset of them would be better) in the proper 
system location when the TeX distribution is installed. Indeed, this is 
what i-Installer on Mac OS X has done for a little while, albeit with 
fonts converted at run-time with fontforge.

Regarding "3.2 Typespec parser". First of all, I need to clarify: by 
"no procedural code allowed" are you saying that you want a scheme like
    \defaultromanfont{Palatino}
rather than
    \renewcommand\rmdefault{ppl}
in the example of choosing fonts?
This sounds like a great idea, and I think some of this may even be 
present in the LaTeX3 project (? see some of the template stuff dealing 
with setting up paragraph styles). It is an idea frequently expounded 
by William F Adams, and it's good to hear another voice in the matter. 
I guess in the beginning, document classes were supposed to be what 
this is all about, but this became complicated by the fact that 
document structure had to be defined there at the same time.

Section 3.4: "The popular pdftex engine should return to outputting 
each page as it is shipped out of TEX’s memory. Whether this is even 
compatible with the production of a PDF output stream, I do not know." 
I think the answer to this question is yes. Witness web-optimised PDFs 
that download a single page at a time in a browser plug-in (perhaps 
introduced after PDF v1.3?). I can only assume that this would work 
with PDF-writing as well.

Regarding turning metafont fonts into OpenType fonts -- I have great 
optimism that this is possible now with one of the various current 
tools (metatype1 and friends) to get to postscript and using fontforge 
to perform the final conversion to OpenType. But it must be understood 
that metafont sources often require massaging before they can be 
successfully converted to postscript. (Unless you're happy to use the 
various tracing tools, which can be successful also.)

I am not knowledgeable to comment on the last sections of the paper 
which deal with XML and getting rid of LaTeX. Does ConTeXt suit your 
purpose? Would LaTeX3? Like I say, I cannot even conjecture what needs 
to be done in this area...

Best regards,
Will Robertson