X-VM-v5-Data: ([nil nil nil nil nil nil nil t nil] [nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil]) Date: Tue, 29 May 90 12:08:07 CET Reply-To: LaTeX-L Mailing list From: PZF5HZ@RUIPC1E.bitnet To: Rainer Schoepf Status: R X-Status: X-Keywords: X-UID: 119 Don Hosek writes: I personally am very wary of any great incompatibilities between version 2.09 of LaTeX and 2.10 of LaTeX. In particular, I think that disabling the old way of doing things, even at the style file level, should be avoided. I agree with him as far as the user interface is concerned. Not only because incompatibilities will pose problems switching from 2.09 to (whatever) but also because of other reasons, see below for details. I don't agree as far as the style file (internal) interface is concerned. While changes to the user interface affect thousends of users style interface changes affect only a limited number of people. We shouldn't judge from our point of view in this regard. Of course style file changes as to be reflected in local style files but at the moment most style files are only variants of the three standard styles and differ only in minor regards. (except for those styles written by the few gurus). Consider, for example, the reluctance expressed by one presenter at last year's TUG conference to use LaTeX internal macros since they seemed to change their function without warning between versions of LaTeX (and despite what Leslie might think, the dates are really not adequate identifiers for what revision LaTeX is on, and I hope that we can incorporate a simple numeric or alphabetic identifier into the version number beginning with 2.10--say 2.10 is the first version, 2.10a after the first release with bug fixes, then 2.10b etc.). The version numbering in LaTeX is surely confusing. One often see somebody telling he is running TeX 2.8 and LaTeX 2.09 and asked what is wrong! With the new LaTeX we should definitly switch to a more transparent scheme. But I don't agree that LaTeX's internals can't be used because they change without warning. They can't be used because nearly nobody understand them. We keep wandering off on tangents in this discussion, which is not bad in itself, but little progress is being made (I'm kind of glad nobody took me up on the quarter bet now). We have to keep wandering off in my opinion and I hope I will see many more mails on this important topic. And anyway, I don't see that we made no progress. I never saw a chance to get a definite syntax by the time of the Texas meeting but I still see a good chance to get to this point in Cork and I hope that a prototype implementation is ready at the end of the this year. By the way Don, if I remember correctly Rainer took you on your bet. People want to know how to do style files in LaTeX, and aside from a specification of Frank's new start section macro, I haven't seen anything concrete towards that end. To make this point clear, the syntax I proposed at that time is *bad* and I hope that we will find something better. It was only a test to check certain things and not a final proposal. The advantages and disadvantages of things like this can only be tested by trying to use it (and so I used it and that's why I don't like it). Users undoubtedly are quite put off by the state of things in the LaTeX world. I've put off installing TeX 3.0 until this week partially because I've been waiting for the new version of lplain.tex that includes \newlanguage and the other necessary definitions for TeX 3.0. I'll spend a little time this week putting together one that checks to see if it's running TeX 3.0 and if so adds the appropriate definitions and send it to this list. If Leslie can officially adopt it, my postings to comp.text.tex on how to install LaTeX under TeX 3.0 can get considerably shorter and there will doubtless be great rejoicing. I did the same and we should look at each others ideas to get an lplain for both TeX 2 and 3 soon. Even if we intend to replace it by something different in the near future. There still is no official word on internationalization. This should definitely be considered for LaTeX 2.10 (has it even been touched on in this group?); TUGboat 10#3 and 11#1 have adequate proof that there is a desire. If you think about agreeing on a standard how to access language specific strings (as done in german.sty babel.sty ILaTeX etc.) it is correct that we haven't discussed this before. I think this is a major topic but I don't like to discuss it at the moment, partly because I think that it will pose not many problems. There has been already so many work at this point so that we need only to collect it and integrate it into the style file interface. Anyway this is a point where mainly the european users are affected so that I would like to defer it until the Cork conference. Back to the attributes: I still vote for a syntax similar to the one proposed by my last message about attributes (please reread it). Reasons: 1) such a syntax is upward compatible which is (not only in my eyes) an important thing in itself. 2) keeping both attributes and optional arguments would be a kludge only if i) the attributes have the same syntax, i.e., look like optional arguments as in Leslies proposals and ii) if the attributes will allow to set/reset everything possible (again as in Leslies proposal). I don't think they should for several reasons. 3) First of all the current set of optionals and star versions give nice shorthand versions because they represent the major attributes (at least nearly always). 4) We also have the cases where it is simply nonsense to change to to an attribute syntax because it would only mean more to type but wouldn't produce a better structure think of \item[gnat] versus the new one \item[text=gnat] or \\*[3pt] versus the new one \item[nobreak|vspace=3pt] or think of the position arguments for tabular and the like would it help to force typeing pos=b ? 5) Explaining the new syntax by is easy and straight forward for new users and easy for old ones if we divide the manual in three or four parts. i) First explain the general structure by telling that every textproducing command in LaTeX is represented by an enviroment like \begin{itemize} ... \end{itemize} \begin{footnote} ...\end{footnote} Explain that these enviroment may take special attributes which will modify slightly the logical meaning of the enviroment, e.g. \begin{section|runninghead=} ... \end{section} ii) In a second step tell the user that for the most common (and short) enviroments (like section, footnote etc.) there exists short versions in command form, i.e., \section{...}, \footnote{...} etc. Also explain that for the most usual attributes there exists posibilities to express them via optional arguments and star forms both for compatibilty and ease of usage. iii) Third, explain how to change certain elements (i.e., introduce visual formatting) by introducing declarations and the like. Also explain that there exists one attribute decls that will take a string of declarations and execute it after the enviroment has set up its default. 6) Of course giving only one decls attribute instead of arbitrary attributes for setting layout parameters will result in some extra typing if this is necessary. But if would make a clear distinction between adhoc formatting and style layout and it would provide all necessary function and would be robust while the other method would lead to chaos in my opinion. 7) One minor point: I pressed on this redesign because I saw the need for something better than the current LaTeX. If the design decisions will be based on `how fast and simple one can make up a new manual out of the old one' and not on what is easier for the user in the final run I think I will stop working on this project. Frank