Let me try to summarize what I understand from this discussion: - LaTeX is not obsolete. Although there alternative formats with various advantages emerging, or used internally in the publishing business, it does not seem that they are ready for general use now, and it does not seem that likely that something to completely replace LaTeX as an authoring tool will be available in the near future. - Publishers do not necessarily use TeX throughout the production process, and would prefer to use other formats especially since publishers want to move into the archiving/database business for which LaTeX is clearly not the ideal format. - LaTeX to XXX converters should be easily in principle, but are tricky because authors mess up documents with all sorts of packages and nontrivial macros. So I conclude what I have been trying to say, maybe not so clearly, before: We need a standard for portable LaTeX which is necessarily a subset of the capabilities of native LaTeX. I think the strongest criterion should be that this standard does not assume that the file is processed through a TeX backend. Also, this seems more or less orthogonal to the goals of LaTeX3, because it is mainly a matter of convention, and not of fundamental hacking in the LaTeX the program. Some comments about things that came up in the discussion: Sebastian Rahtz wrote: > how about separating typesetting from editing and manipulation? its > not the only paradigm [...] > i do have one last (trite) remark - why are you (the academic authors > reading this) spending so much time on arguing about typesetting and > publishing? why don't you spend my tax money on doing research into > your subject......? It is important to keep in mind that most of the documents that we academics write don't go via a publisher. These are class notes, informal exchanges, short reports, grant proposals etc. Most of these are routine, but some are important and significant documents. Thus, the typesetting side is of crucial importance to everyday academic life; separating typesetting from authoring is therefore not an option in the majority of instances. Moreover, thanks to LaTeX, we can typeset even routine documents to a high standard (this is not to say that it's impossible to screw up badly). It is therefore frustrating to see that in many cases when the manuscript hits the publisher, and is copy-edited by a professional, things get worse rather than better. When manuscripts are retyped (and most publishers try their very best to do this carefully), it always introduces new typos. When publishers take the LaTeX file, some choke on vanilla amsmath (although it's explicitly allowed in their author instructions) and other problems. In short, the fact that most publishers cannot accept a carefully prepared LaTeX file causes hours of proofreading on a level which is below the standard of the submitted document. We have every interest to avoid that, and I think that's the same for the publishing professions... Hans Aberg wrote: > I think this will affect also scientific publishing: A lot of > scientific results can be better presented using multimedia. Even in > pure math, one True, but I think this is something that is very far from the areas where LaTeX is the optimal engine, and a unifying approach would be even more utopic, so I think this should not be of immediate concern here. "Y&Y, Inc." wrote: > No. You can't review articles. You can't provide easy searching and > access to all articles in an area. Etc. There *is* a role for > publishers to play. Of course it is true that there have always been In my experience most of the work in the review process is carried out in academic institutions. Very often papers that I review are sent by the editor who is naturally part of an academic institution through the mail system of this very institution, where only the letterhead refers to the non-academic publisher. Re: figures and images: Sebastian Rahtz wrote: > i know i sound like an evangelist, but XML/MathML/SVG really *are* > designed to cover this sort of game. your SVG graphic will embed > MathML markup cleanly. Again, are the necessary authoring tools available? Will it allow me to easily typeset (!) my personal documents? Thinking about psfrag, it seems that its use is not only problematic when portability of other formats is required, but psfrag also seems to be a potential problem within the LaTeX paradigm: Change of document classes, font size etc. may affect the embedded graphic badly (depending on the type of material). On the other hand, I do see the usefulness of the functionality that psfrag provides. Would it be hard to write a script which takes an eps file, runs it through LaTeX/psfrag, and converts the dvi output back into eps with the same bounding box (preferably not using bitmapped fonts and including only those fonts that are needed)? This way one could meet stricter portability requirements with little extra work for those cases where they are needed. Generally, such issues need to be addressed. Maybe even Elsevier would be more friendly towards accepting LaTeX as an exchange format if the success rate for conversion into XXX could be significantly improved. Maybe one could even try to implement the equivalent of -Wall into the LaTeX engine (or as a package) so that authors could check without pain if their documents could mess up non-TeX backends. (I guess that document conversion is not just an issue for submission of scientific articles, but also that almost everybody will at some point have to convert something into HTML, Word, etc.) If portability standards for LaTeX can be defined, maybe one can introduce a compliance tag into the package definition interface where the package signals whether it's non-TeX-backend-compliant (maybe with some gradation; this is vague, but I am not sure about all the issues potentially involved). If someone now says why not SGML then: The advantage of LaTeX from the author's point of view is that it is a single platform for authoring, typesetting and document exchange (where, I believe, the problems can in principle be minimized to ensure that LaTeX adequate for all practical purposes). Marcel