From buckley@agb.RoyalRoads.ca Sat Aug 29 02:24:57 1992 Received: from d3.RoyalRoads.ca by dkuug.dk via EUnet with SMTP (5.64+/8+bit/IDA-1.2.8) id AA14838; Sat, 29 Aug 92 02:24:57 +0200 Received: from agb.RoyalRoads.ca by Post.RoyalRoads.ca (5.61/5.17) id AA23934; Fri, 28 Aug 92 16:20:14 -0800 Received: from localhost by agb.RoyalRoads.ca (4.1/1.34) id AA08826; Fri, 28 Aug 92 17:25:08 PDT Message-Id: <9208290025.AA08826@agb.RoyalRoads.ca> To: SC22 WG5 Subject: document processing Date: Fri, 28 Aug 92 17:25:07 -0700 From: buckley@agb.RoyalRoads.ca X-Charset: ASCII X-Char-Esc: 29 I have been reading the traffic, or at least that portion of it kindly distributed to people not on the X3J3 mailing list, regarding "Processing Words" or "Word Processing" with some interest, particularly since I was not overly happy with the results of the Fortran 90 effort in that regard. I notice comments like "use fmgrep" in some of the mail. I pick this only to illustrate one point: some form of 'universality' should be a requirement. I do not have fmgrep; I assume others may not. I presume, but do not know, that it is related to grep, egrep, and 'anyother*grep' you may wish to list. Whatever is done should, to as large an extent as possible, be accessible to everybody. What I think is completely lacking in the discussion is a set of objectives. We want an electronic form of the document; presumably that has been decided. We want to be able to distribute that document, but how widely? That does not appear to have been decided. I have seen some comments regarding this point, but if we first decide on who should have access to the document, and for what purpose, it will certainly make it easier to decide what form the document should have. Once we decide who should have access to the document and why, we can discuss what requirements there are for an appropriate text processing systme. I would like to suggest we first discuss the requirements. Then we should pick the system which most closely fits the requirements. I for one would have to drown my tears in a bucket of beer if Word or one of its compatriots was chosen: I happen to have much more belief in the acyonym WYSIAYG. If by chance someone doesn't happen to have seen it before, that means 'what you see is ALL you get'. But in fact, even though it is certainly not my choice, if we set out a list of criteria, and Word fit the bill best, then I would not have any objection to its choice. (This is not a complaint specifically about Word, but about the so called WYSIWYG PC-Mac type word processors in general; I actually use WriteNow, and the same comments apply. I routinely and quite happily use WriteNow for many things, but as they say---please forgive me, ladies---get a man to do a man's job: I would never use WriteNow or anything of that ilk for a paper I am submitting for publication.) I would like to suggest the following as a preliminary list of requirements. [I have added some personal comments to some of them.] I suggest we discuss this list, or some similar list. What criteria should we use in choosing a text/word processor? I would be quite happy to see someone else suggest a different list, or to suggest modifications to the one I propose, but at least let us discuss what issues are important. Note that I am going to make one assumption which I feel is absolutely crucial: there will be one master copy of this document, maintained in one place, probably by one individual (the editor), or at least by a very small group. Copies would be regularly distributed, or publically available, primarily so that members of both committees would be able to have access to recent, up-to-date copies of the document which they could print, preview, or process, as they wished. Corrections, amendments, or proposals for change would only be given as edits directed at the current master copy; only the editor would actually modify the master. If may be convenient if submitters can send material to the editor in a form ready for direct inclusion into the document--the editor may then choose to do so or not, at his discretion--but I do not view this as a requirement. On the other hand, based on experience with developing the original L12 document, and other experience, I do not want to see multiple authors distributing 'modified' versions of some or all of the document; figuring out what the changes are is then impossible. Here is a suggested set of requirements for discussion: printability: The document should be in a form so that as wide a group of J3 and WG5 members can use their own facilities to print the document. [I will assume here that we require access to at least a 300 dpi laser printer or the equivalent; it is silly to pretend we can output this document on a typewriter or its equivalent.] invariability: Anybody who receives the document and prints it on his (or her of course) facilities should see the *same* document. The line breaks, page breaks, and so on should be identical. The quality of the output may vary with the device, but not the layout. [This must also apply to the *final* printed document; we must not be told, as we were with Fortran 90 that "yes, that is poorly typeset, but on the final document, it will be fixed". I want to see exactly what we are getting.] transmitability: It must be absolutely straightforward to send the document from one site to another. This almost certainly means that it must be possible to sent it in a straight ASCII form, and that should not depend on utilities available on any particular systems. [For example, one can do it with btoa and atob on Unix, but that is unacceptable since these are not universal utilities.] universality: [As many of you may know, this one is dear to the hearts of us Canadians.] As much as is feasible, the system must not be tied to any particular vendor or system. One should be able to receive, process and print the document on a wide variety of machines. [I immediately think of PC's running DOS, OS/2 or Windows, Macs, Unix boxes of a semi-infinite but never quite the same variety, VMS and CMS; but others no doubt have a different list, and hopefully our choice of document system will accomodate possibly different systems for J3 or WG5 members who join the process later. A small list of possible systems should not determine our choice.] reasonable cost: This factor should not be ignored. I am concerned when I see references to Frame. I do not have Frame, and probably cannot get funds to purchase it, since it does not happen to be the system which DND in Canada has chosen. And I view offers from vendors to provide free copies, although certainly generous, with a certain amount of hesitation, especially when the suggestion is that a small number of copies would be given to selected individuals. [Almost certainly, if any reasonably expensive product is chosen, access will be a problem for someone, if not me personally.] variety of capabilities: The system must have a wide variety of capabilities as an *intrinsic* part of its design. A partial list would include: o a wide choice of font styles and sizes o automatic numbering of entities, e.g. equations, syntax rules, sections, and so on. In conjuction with this, references to numbered quantities must be capable of being automatically generated. [I was astounded when I discovered that with the Fortran 90 document, if sections were changed, then all affected section references had to be *manually* changed. We are still seeing editorial corrections to fix some of these.] o automatic index and table of contents generation should be available. Thus there would be an index with every copy, not just the final version. o automatic bibliography generation should be possible, much as for the last two points. o global changes in the document should be possible. For example, if it is decided to change the font type for all sections headings, or for all deprecated text, or to change the font size for all footnotes or all subsection headings, it should be easy within the capabilities of the processor. Making global changes manually is unacceptable, and relying on awk, sed or the like in this regard is a poor substitute for a capability which should be within the processor. o capabilities for reorganizing text should be available; one example noted was the ability to automatically generate a list of syntax rules for an appendix. o facilities for building complex tables, setting equations with possibly complex mathematics, and for dealing with figures must be included. o capabilities for control of typographical difficulties should be built in; e.g. having complex font changes, or avoiding widow lines, bad hypenation, or poor page breaks. viewability: It should be possible to view the document page by page on a PC or workstation screen. This of course comes part and parcel with the WYSIWYG processors like Word, but it is highly desirable if a markup language is used. [Of course, many systems do allow this, as for example with Postscript files, or TeX or LaTeX.] processability: The document must be in a form which facilitates processing. In this regard, we do not need universality. Each individual may do with his copy of the document as he wishes. What he does is not in need of being repeatable on all platforms onto which the document may be copied. But the key point is that the document form must be amenable to such processing. [That tends to be more difficult in documents where the file which is stored has 'hidden' formatting information, as in the typical WYSIWYG system.] Those are some of the criteria which I view as important. In fact, the list is probably a good approximation to my view of the relative importance of the various points, in decreasing order. Let me finish then with a few opinions of my own. I have my own personal bias of course, but that is because it fits my view of what I feel our requirements are. Just so that you know what my bias is, I will tell you that if the choice were strictly mine, I would come down solidly for LaTex. On the other hand, I am only peripherally familiar with products such as PageMake or Frame, so I don't pretend my choice is necessarily correct. I am however very familiar with LaTeX and Tex, and would feel very confident that they could do the job. Frame: NO Why? Looking at Jerry's comments, it fails badly on the cost issue and the universality requirement. Yes, we do not all need to be able to process the document, but yes, we do all want to be able to quickly ftp (or something) the latest copy and print it. I do not at the moment see how that would be possible. Word: NO Why? I think it would fare poorly on the set of built in capabilities, but since I am not a Word expert, I do not want to say this very strongly. But my experience with Word, WordPerfect and WriteNow indicates that though they do some things really well, they can not come close to approaching the enormous capabilities available with something like TeX or LaTex. I find is almost amusing that Word somehow comes with a 'Basic interpreter'--I remember such a reference, but have no idea of what it is. Perhaps it is a way of adding a markup language to Word? --in which case, why not go with a real markup language. SGML: ? I can say little about this, since quite frankly it is a subject on which I must admit complete ignorance. But since it is a markup language, I think it must be put head to head with TeX (or LaTex, from a practical point of view). TeX is widely used, and, whatever one may view as its faults, it is certainly accepted by many as having enormous capabilities. One may point to many examples to demonstrate the quality and versality available with TeX. Wordperfect: God forbid. I see enough of it at work, thank you very much. LaTeX: Yes As I indicated, this would be my choice. I would add though that I would expect that someone well versed in TeX itself would be needed; there would be some situations where the macros of LaTeX would be inadequate, but where the underlying capabilities of TeX would do the job easily. LaTeX is after all just a macro package built on TeX. This is mostly how I use LaTeX anyway. Keith indicated that TeX and troff were beyond his pain threshold. In view of my comments above, that would be irrelevant, unless he choose to be on the editorial committee. What most of us would receive would be a straight ascii TeX document; we would 'TeX' it, and then preview it or print it as we wished. As for Keith's comments that 'any' word processor of any stature handles numbering and so on well--well, that would require some convincing for me to accept. Postscript: No Although this would solve the cost, printing and transmission problems, it would fail badly on being able to process the document for information. It is not a form that anybody could work with in any meaningful way. Anyway, that is my two-bits (should I say ten-bucks?) worth... and Greetings to all from Beautiful B.C.---and after the recent meeting, I hope I don't get any argument on that point! Bert