52

I was just browsing the new questions and noticed egreg's recent answer, in which he uses a certain extension from e-TeX and cautions the reader that it is not compatible with Knuth TeX. And so I wondered if this was a real restriction in practice, which does not seem to be answered here, though according to Frank Mittelbach's answer, e-TeX is expected to be available to everyone. Certainly it's been around long enough; hearing "but you need e-TeX" sounds to my ear like "but you need a 386 processor or higher". I can imagine that some hidebound publishers might actually have not updated their TeX in 20 years, and so can Donald Knuth, but why make this kind of allowance in general? Is it because the TeXbook, the standard textbook, does not describe it?

Edit: After barbara beeton's comments and Mike Renfro's answer, I see the need to clarify a bit. My question concerns:

  • Notable individuals, or notable groups of individuals, who, for some principled reason, do not have e-TeX support in their installations. In this case, I'm curious how this came about.

  • Institutions or distributions of some kind that do not allow the use of e-TeX by their users, whether or not those users have any particular opinion on the matter. Again, I am curious why.

Basically, to what extent is e-TeX the default, and what's the deal with the exceptions? The second point actually excludes the AMS, since as barbara says, they actually go to great lengths to maintain consistency despite updating their system (though I would like to know if any new papers or books are produced under conditions that exclude e-TeX). But the ACS is an example that Joseph Wright explained well. And in the first point, I suppose Knuth himself (barbara's answer) is an example, as is the PGF project (percusse's comment), though why the PGF project excludes e-TeX is sort of begging the question, since it seems to be to support users that don't use e-TeX.

Community
  • 1
Ryan Reich
  • 37,958
  • If I'm not mistaken, author and the maintainers of TikZ/PGF have e-TeX-free code policy (presumably to be compatible with plain TeX). – percusse Jun 16 '12 at 17:50
  • I have found where I got that info. Please read the comments of @cjorssen to the second answer of Can pgfkeys deal with active comma. – percusse Jun 16 '12 at 19:08
  • 3
    Regarding the note from Knuth about the font versions: As I understand it, the problem was not exactly that people were not updating their metafont versions of computer modern, but rather that most people used the postscript type 1 versions of the fonts, which were not updated to incorporate all of Knuth's 1992 changes until quite recently (2008 I think). – Lev Bishop Jun 17 '12 at 04:14
  • @Lev: That's a very strange story. You are saying that the fault for this extreme delay in upgrading lies with whoever generates the postscript fonts, and not with the people who (presumably generally) used whatever version was provided? What happened there? – Ryan Reich Jun 17 '12 at 05:00
  • 4
    my understanding is that generating good (well-hinted, etc) type 1 postscript fonts is very difficult and that there are very few people who can do it well, those people rarely volunteer their services for free, and organizations like the AMS (who have mostly been the ones doing this) have limited funds for such things. A few more details in the amsfonts readme, here. barbara beeton probably has more insight. – Lev Bishop Jun 18 '12 at 15:15

4 Answers4

31

A lot of server-based set ups are very conservative. For example, I write achemso to support submissions to the American Chemical Society. On the servers they have to take author .tex files and produce .pdf files, the e-TeX extensions are not available. (I am told an upgrade is planned for later this year.) These systems are often based on custom additions to the basic code, and have to be absolutely stable. So it is unsurprising the extensions are not available.

A second area is people who use plain TeX. One of the reasons for doing this is knowing that the binary is not going to change, so files written many years ago can still be processed to give identical input today as they did when written. Any change in the binary can affect that, and so for absolute identical output those authors will use Knuth's TeX, not e-TeX, pdfTeX or other derivative which contains the extensions.

Joseph Wright
  • 259,911
  • 34
  • 706
  • 1,036
  • 4
    By "absolutely stable" you mean that they are reluctant to upgrade and thus risk breaking their system, not that they fear that e-TeX itself is unstable software? – Ryan Reich Jun 16 '12 at 16:06
  • @RyanReich I would assume so: at least in the case of the ACS there is little TeX expertise, they just 'buy in' a system. – Joseph Wright Jun 16 '12 at 16:09
  • 7
    @RyanReich -- production systems that handle large quantities of material, some of which has a significant delay between accession and publication, must be absolutely stable, to avoid, for example, having to re-proofread something that was already "final" and simply awaits author confirmation (which can sometimes take months). at the ams, the version of every binary and package used is recorded, and if a rerun is necessary, the same versions are used. these same versions are used again if an article is republished in a collection or a new edition of a book is prepared, sometimes years later. – barbara beeton Jun 16 '12 at 17:39
  • @barbarabeeton I'm intrigued by one part of this: the idea of retypesetting material. At least in my area (chemistry), we get galley proofs and have only a few days to get them back to the publisher before final production. So there is at most a couple of months when a document has to be typeset. After that, they archive the PDF, and there will never be a need to retypeset. What are the cases you see in your area where this happens? – Joseph Wright Jun 16 '12 at 17:44
  • 5
    @JosephWright -- the policy here is that the (la)tex source and all graphics and author-supplied packages are archived along with the pdf. if a reprint is needed, and there are no corrections, the pdf is used, but if there are corrections (talking about books here), the tex job is rerun. using the same versions of everything ensures that only the changes (and whatever flows "downhill" in a chapter) need to be re-proofread, although a cursory page check is made elsewhere. math has a long "shelf life". – barbara beeton Jun 16 '12 at 18:10
  • 2
    @barbarabeeton: But the AMS does allow e-TeX in new documents, right? Is there any situation where e-TeX would not be available to a new paper? – Ryan Reich Jun 16 '12 at 19:02
  • 3
    @RyanReich -- the ams is currently using tex live 2010 "out of the box" (along with a local tree) for production, so yes, e-tex is available. we intend to investigate switching to tex live 2012 when that is released, in order to stay not too far behind what new authors are using "outside". (the last switch took almost two years for material in the queue to "catch up" by itself, with a final forced switch/reprocessing/rechecking of a few dozen article stragglers; for books, the original version was retained.) however, for articles, very little e-tex should be needed by authors. – barbara beeton Jun 17 '12 at 12:47
  • 1
    I'm accepting this answer (and the comments beneath it) as likely accounting for most non-eTeXers. – Ryan Reich Jun 22 '12 at 06:23
17

if a bug is reported in tex, knuth will accept only a minimal example processed with the latest unmodified version of tex. (of course, a bug elsewhere would never be considered for submission to knuth.)

barbara beeton
  • 88,848
5

PCTeX v6, at least not by default.

Mike Renfro
  • 20,550
4

Unless there is newer information, Scientific Workplace does not appear to support e-tex.

Steven B. Segletes
  • 237,551