On 15-Mar-03 Alexis Lee wrote:

On Fri, Mar 14, 2003 at 07:28:28PM -0000, Ted Harding wrote:

...

-- it covers the use of Unix troff and friends, and the use

Thanks for this Ted.

If people are looking at text-based document formatting systems, LaTeX is usually considered superior to troff &c. For print only of course.

Particular strengths are *really* nice equation formatting, automatic table of contents generation and automatic bibliography generation.

A fair, but arguable, comment! Granted: TeX/LaTeX has the benefit of coming later on the scene; has more typesetting wisdom built in (Knuth being a guru on that as well as other things); does usually produce nicer-looking output, espcially for mathematics.

However, much of its "nice-looking equations" is due to the great variety of well-designed fonts and special symbols which Knuth had the wisdom to create. Now that these are available as PostScript font definitions, which can be installed in groff as well, this advantage is not intrinsic to TeX. Indeed, given any printed output from TeX, you could in principle exactly duplicate in in groff (since this permits you to place any mark whatever, so long as you can define it to the program, anywhere on a page), and in practice reproduce it at least very closely if not identically.

The equation formatting from troff/groff is also very good, and can be tweaked for better effect if you wish. The main difference in most practice, as noted above, lies in the fonts used.

There are features where I would back troff/groff to produce better output, for the same effort, than TeX (though again, since TeX can be "programmed" by the user, in theory it too could do the same).

Automatic ToC, indexing, and bibliography gerenation are equally within the scope of troff/groff (though best results are obtained by using non-groff auxiliary programs such as Tim Budd's 'bib' and, for indexing, ironically the TeX utility 'makeindex' which can eqwually well be set up for troff/groff). For ToC, the best method involves a simple 'awk' script (see the source tarball for "Unix Text Processing" whose URL I gace previously).

In terms of capability, therefore, I would not rank TeX as generally better than groff. On the other hand, it is a fact of life that in the academic and technical authoring world, TeX is far more widely used. This is largely because it was adopted with enthusiasm when it first became generally available around 1990, resulting in development of good "wrappers" such as LaTeX and the writing of much excellent documentation; at that time, troff was still hard to use and sparsely documnted. So it is the "industry standard", if you like, espcially for document exchange in that world. You would be cutting yourself off if you worked in that world and did not want to have anything to do with TeX.

On the other hand, as a certain widely used word processor exemplifies, being the "industry standard" which "everyone" uses and which you have to use youself if you are to have dealings with them, is not of itself an argument for the _quality_ of that software. Not that I'm hinting that the quality of TeX can be measured by the quality of that other thing -- quite the contrary!

Best wishes, Ted.

-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@nessie.mcc.ac.uk Fax-to-email: +44 (0)870 167 1972 Date: 15-Mar-03 Time: 13:53:48 ------------------------------ XFMail ------------------------------

On 17-Mar-03 Adrian F. Clark wrote:

(Ted Harding) writes:

...

In terms of capability, therefore, I would not rank TeX as generally better than groff. On the other hand, it is a fact of life that in the academic and technical authoring world, TeX is far more widely used. This is largely because it was adopted with enthusiasm when it first became generally available around 1990, resulting in development of good "wrappers" such as LaTeX and the writing of much excellent documentation; at that time, troff was still hard to use and sparsely documnted. So it is the "industry standard", if you like, espcially for document exchange in that world. You would be cutting yourself off if you worked in that world and did not want to have anything to do with TeX.

I'm afraid I have to rise to this. TeX dates back to around 1978, not much longer after the emergence of n/troff. Knuth performed a major re-write in 1982, changing the original Pascal to his (still Pascal-based) Web system. LaTeX actually dates back to 1984: I still have pre-release copy of Lamport's book dated the following year.

You are of course right about the early history of TeX; but I was talking about "became generally available" -- perhaps I should have said "began to be widely used". I don't recall seeing TeX being used by many people at all until 1988-89; by 1990, however, usage had spread widely.

Unix roff dates from the 1970 port of runoff to Unix on the PDP-11, and was "upgraded" to nroff ("new roff") by 1974; troff ("typesetter roff"), essentially the same program but enhanced to cope with the much finer rendering available from phototypesetters, was developed in 1975, and did not need much tweaking to deal with laser printers (such as Apple's) which began to come out around 1980. I would say that n/troff, amongst those people concerned with computer typesetting (especially complex or "difficult" stuff like mathematics), was very widely used by 1980.

The reason that TeX caught on while n/troff didn't was that TeX was actually much easier to port: it may seem hard to believe now but Pascal compilers were much more widely available than C ones in the early 1980s, and the n/troff sources were restricted by AT&T's licenses to boot. I know because I ported TeX to VAX/VMS in 1982 so that I could use it for my PhD thesis. (I still remember my first page of output, obtained by putting dots onto a sheet of paper with a pen plotter!)

Those are interesting comments. And a valiant effort for your thesis! (I suppose you found other means in the end, but I boggle at the idea of a whole thesis reastered by a pen plotter!).

The principal reason for the continuing popularity of (La)TeX is certainly the size of the development community; however, another major factor is that the TeX engine is almost totally bug-free: the only bug report discussed on the TeX-implementers' mailing list in the last couple of years is concerned with i8n issues and the trip test, which is certainly esoteric.

The only place that TeX falls short of n/troff+tbl and hence groff is in the setting of tables, in particular producing entries that span several rows. The pic and graph pre-processors are also more elegant solutions to some forms of graphical typesetting than the TeX-oriented equivalents.

Not forgetting that 'pic', as well as implementing a diagram description language (a sort of sophisticated LOGO), also incorporates a powerful numerical computation capability, as well as looping and branching constructs. You can therefore incorporate into your document source 'pic' statements which will read data from a file, perform computations on the data, and determine what form of graph shall be drawn and what annotations should appear on it (e.g. plotting a set of x-y data, estimating a straight-line fit, plotting the line, and labelling it with its equation; at the same time, the components of variance that would go into an analysis-of-variance table for the data can be computed, and made available to incorporate into a following Analysis of Variance Table formatted using 'tbl'; and all this on-the-fly while actually typesetting the document!).

Another feature of n/troff which I'm not aware of being part of TeX (could be wrong here and would welcome correction) is the possibility of planting a request into the document source to run an arbitrary command externally; that way, in the course of formatting the document, some really nasty job can be farmed out to the operating system and its results read back in to the document. Mind you, this is definitely Unix-think!

Having said all this, I applaud your group's efforts in entering the book text: I'd hate to see troff disappear totally. Remember: the main advantage of the Unix approach is that it gives us all choice.

Thanks! And while troff my be something of a dinosaur, I don't think it's in much danger of extinction at present. It's much more than a man-page formatter.

Ted.

-------------------------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@nessie.mcc.ac.uk Fax-to-email: +44 (0)870 167 1972 Date: 17-Mar-03 Time: 16:24:29 ------------------------------ XFMail ------------------------------