rcd@ico.isc.com (Dick Dunn) (02/21/90)
Here's the first piece of the results from the troff survey I posted three weeks ago. There will be one or two more parts, depending on how I carve up the details. Quite conveniently, there were 102 responses, of which two contained only selected comments, leaving just 100 actual tally-able responses. Please realize that this survey was *not* controlled in any useful way. Therefore it's wrong to attempt any serious or detailed statistics on the results; however, I feel that some broad general conclusions are possible. In order to allow easy interpretation (and discourage overanalysis of numbers:-) I'm showing the results in a rough bar-graph form, with a ruler at the top to show %age by 10s, thus: +---+---+---+---+---+---+---+---+---+---+ mumble |*************** phumpher |******** This was for 37% mumble, 19% phumpher. Further interpretation: If people have additional questions about the results, I'll try to answer them *as long as* they don't require re-analy- zing the data. That is, I can probably answer things like "Did you see x?" but not "Is there a correlation between abc and xyz?" Each of the questions is identified by its original number and a synopsis of the question. I've rearranged a bit for convenience. After the graph I've added some interpretation and shameless editorializing. (1) percentage of work time spent writing/modifying documents: +---+---+---+---+---+---+---+---+---+---+ 0-10 |********** 11-25 |************* 26-50 |************* >50 |**** I divided the results like this because the data suggested it. There's a category of occasional work, up to 10%. Then there are folks who spend a goodly chunk of time now and then. Next are the people who, from notes and other comments, spend a lot of time on documentation although it's not their only job. The last group is really "writers." The distribution suggests that the cross-section of people who answered is a useful one. (2) currently use troff? +---+---+---+---+---+---+---+---+---+---+ not_at_all |**** a_little |***** some |****** a_lot |************************ OK, we've got a lot of troff users...but we've also got a useful handful of people who are using something else in-addition-to / instead-of troff. So, what, for instance, do they use?... (3) other formatter(s) used? +---+---+---+---+---+---+---+---+---+---+ TeX |************** uSoft_Word |****** Word_Perfect |**** FrameMaker |*** Interleaf |** Ventura_Pub |** Word_Star |** Scribe |* (Note that multiple answers were possible, and common. One that came up reasonably often was TeX + some PC-based formatter.) A couple of interesting observations here. First, TeX is the only serious contender for this group of people. Second, consider that neither troff nor TeX are wysiwyg. (More about that later on.) Scribe's rather poor showing may have more to do with cost and marketing than technical merit; some folks would use Scribe if they could afford (or even obtain) it. There were various other formatters represented, in onesy-twosies. (4) % of your documents use troff (IF other formatters used)? +---+---+---+---+---+---+---+---+---+---+ <10 |******** 10-49 |**** 50-74 |**** >=75 |********** This shows an interesting split. Of the people using other formatters, some have essentially abandoned troff; some still use it heavily. The "hole in the middle" suggests that people are either holding on to it or getting rid of it, but having two or more formatters on relatively equal terms may not be a happy situation. (8) importance of troff? +---+---+---+---+---+---+---+---+---+---+ none |** existing |***** some |**** most/all |*************************** Expanding the category explanations, they are "no importance", "for existing docs only", "ongoing for some", and "ongoing for most/all". This question is interesting as compared to the use of other formatters: Troff is not being abandoned just because some other formatter is available. (6) preprocessors used? +---+---+---+---+---+---+---+---+---+---+ tbl |************************************ eqn |************************** pic |********************** grap |********* refer |*** psfig |*** A few others were mentioned (bib, fig, some local ones) but down in the noise. The 90% use of tbl is interesting. (It does suggest that people who go mucking about with the preprocessors ought to think twice about whether they've broken anything!) Overall in the survey, the preprocessors tended to get good marks and are obviously considered an important part of the whole package of documentation tools. There is some weighting based on the fact that tbl and eqn have "always" been available (since the earliest generally distributed versions of UNIX), but pic was introduced quite a bit later and grap only relatively recently. Still, note that tbl stands out even against eqn. (7) macro packages used? +---+---+---+---+---+---+---+---+---+---+ mm |***************** me |********** ms |******************** man |***************************** mv |* local |****************** Given the tendency to tinker with macro packages (among vendors as well as individual sites), the above data suggest "Please don't break `man' when you're messing around with it!" Other points: The number of "local" macro packages is interesting. It is somewhat inflated by the fact that some people consider major modifications to an existing package to be a new local package--fair enough, as long as you don't count all of these as done from scratch. Also, the preference for ms over mm is (to my thinking, such as it is:-) strange. I suspect it is due to history and availability: mm is not distributed with older systems, including BSD. (9) trouble sources? +---+---+---+---+---+---+---+---+---+---+ troff |************ macros |****************** preprocessors|******** I had asked for which one of the three areas was the major source of trouble, if any. However, I tallied answers in which two or all three were cited. Obviously, the macros stand out. I'll have more to say about troff as a source of trouble in the next posting. For preprocessors, from looking at other comments, pic is the biggest source of trouble. (Pic *is* considered valuable; it just seems to be hard to use and frequently counterintuitive. Some folks have tools to generate pic input.) _ _ _ _ _ A note about wysiwyg (what you see is what you get): One of the issues I was watching for was this: Is wysiwyg a big deal? How much do people want it? To what extent are they leaving troff to go to wysiwyg formatters? The answer, at least for the group of people who answered this, is "not much". Wysiwyg is a big yawn, as far as I can tell. Quality of output, programmability, flexibility, and control are more important. I went looking for comments which mentioned wysiwyg explicitly or implicitly-but-obviously. I counted 1/2 for an indirect reference and 1 for a direct ("I [don't] want it") reference. The final count surprises me: It's about 6 units for wysiwyg and 9 against (out of 100). I wouldn't interpret this as an anti-wysiwyg sentiment; it's more accurate (given the statistical vagaries) to say that this group is as likely not to want it as to want it. However, many people mentioned previewers. That was the one question I really missed asking. Even folks who explicitly want a non-wysiwyg, markup-oriented edit-format split would like a previewer. -- Dick Dunn rcd@ico.isc.com uucp: {ncar,nbires}!ico!rcd (303)449-2870 ...Don't lend your hand to raise no flag atop no ship of fools.