mckeeman@wivax.UUCP (William McKeeman) (10/31/83)
About 1/2 of the users use emacs, and about 1/2 use vi. pmail is the main consumer of emacs cycles. Heavy users of editing either use vi because reliabilty and speed are more important that function, or use emacs for the extension language and modeless editing. emacs is hard to learn. The key mapping abilities make it difficult for one user to help another, you can never tell what key is bound to what. vi is a pain to use. so, you take your pick...
mem@sii.UUCP (Mark Mallett) (11/01/83)
b On the VMS (VAX 11/780) system that I use, almost all of the users came from a DEC-20 and there is a widespread sentiment that "I'd love to use EMACS if only we had it". Well, we do have it; but users can run it only if they manage to discover its presence serendipitously. The system manager has, probably rightly, elected NOT to publicize its existence because it is NOT supported, nobody knows much about it (including where it came from; ergo where to get assistance from), and mostly because it has abysmal documentation. I realize that this doesn't relate to "emacs vs. vi". Btw.. what about vms emacs? anybody out there have any involvement with it? Myself? I use emacs. But then again, the alternative is EDT. Mark E. Mallett decvax!sii!mem
tjt@kobold.UUCP (T.J.Teixeira) (11/01/83)
This is in reference to Steve Zimmerman's comment on editor usage at
CCA, and in particular the comment:
Gosling's EMACS is available on both VAXes, but nobody uses it.
It is very seldom that an editor gets to be the dominant editor on a
system because all the users have tried all the alternatives and made
their choices accordingly. Rather, most people will use the editor
that is best supported. If you go to your local guru and ask a
question about vi, and he answers: "I use EMACS, not vi! You'll have to
figure that out yourself.", how long will you continue using vi?
On system which have sizable user communities for different editors,
the editor usage usually closely matches some other division within a
larger organization. For example, on some of the PDP-10's at MIT, the
CLU people tend to use TED (a screen editor written in CLU) rather than
EMACS. On one of the VAX's at MIT, most people use Gosling's EMACS,
even though an older version of CCA EMACS is available as well as EE,
another local variant of EMACS. This is true even though CCA EMACS and
EE are closer to PDP-10 EMACS. Why? Partially, for technical reasons,
but mostly for historical -- we got Gosling's EMACS first, and the
system maintainers and older users know it. That is what is introduced
to new users. The community of users for EE mainly consists of people
who also use either a PDP-11 or the MIT-NU terminals (a 68000 system).
This reflects the fact that Gosling's EMACS didn't run very well on
these systems (at least when it was first tried), but EE does. These
people use EE on the VAX because then they can use the same editor on
all their machines.
In all this discussion, I'm reminded of the time that DEC gave a talk
at MIT when the VAX-11/780 was announced. One of the first questions
was whether TECO would be available. The response:
Text editors are like religons --
only people are more fanatical about text editors!
--
Tom Teixeira, Massachusetts Computer Corporation. Littleton MA
...!{harpo,decvax,ucbcad,tektronix}!masscomp!tjt (617) 486-9581z@cca.UUCP (Steve Zimmerman) (11/02/83)
Tom Teixeira says "Most people will use the editor that is best
supported." I fully agree. However, the implication is that
significantly more effort is put into supporting CCA EMACS at CCA than
is put into supporting other editors. This is not the case. Very
little active support is put into any of the editors in terms of helping
users on a one-to-one basis. Instead, it is the support that comes with
the editor that is important. Both the Rand editor and vi come with
reasonable documentation, which is one reason they both developed a fair
following. As CCA EMACS is a much more complex editor than either of
those two, its documentation is correspondingly more extensive,
including a 235 page manual, an online tutorial that assumes no previous
knowledge of text editors, and many other online help facilities.
Certainly this type of support was instrumental in the acceptance of CCA
EMACS here. Without it, there would have been no way that CCA EMACS
could have been introduced as the standard editor for secretaries,
managers, and our text editing department. We do not have the
programming support to offer a lot of interactive help to these users,
and these people typically do not like to have to look through source
code to figure out how a a certain feature works.
Correspondingly, the lack of this level of documentation was certainly
one reason why Gosling's Emacs never caught on here. I have a copy of
his manual; several users even borrowed it to look at, but none began
using his Emacs as a result. When Unipress announced with great
fanfare that they were marketing Gosling's Emacs, and word got out that
they were planning to do a real manual, I sent away $30 to see what the
new version was like. I was quite surprised to find that the manual was
just a slightly updated version of the Xeroxed sheets I had, now put
into a three ring binder. In addition to finding the already noted
spelling and grammatical errors, I was struck by the incompleteness of
the manual. For example, the entire contents of section 19.2, entitled
"Electric-lisp-mode -- Assistance for Lisp programming" is the phrase
"No documentation yet".
Tom is also correct when he mentions the phenomenon whereby people tend
to stick with the editor they learn first. This is undoubtedly another
reason why we have no users of Gosling's Emacs here. But at the same
time, the fact that most users of other editors at CCA have switched to
CCA EMACS seems to indicate that this resistance can be overcome when
the difference between editors is great enough.
Of course, there are many technical reasons why people at CCA prefer CCA
EMACS to Gosling's. However, I must agree with Tom's comment that people
are more fanatical about editors than religion. (How many people here
remember the 1980 Usenix conference in Deleware where Dave Yost started
his talk on the Rand editor with the statement "Friends, be saved!")
So, I think I had better stop here before I ignite too many flames.
Steve Zimmerman
{decvax,linus}!cca!zwarren@ihnss.UUCP (Warren Montgomery) (11/03/83)
I didn't see the origin of this discussion, but I have a lot of data on editor use at BTL, collected from examining accounting files, rather than any personal knowledge of the users. It certainly isn't a scientific study, nor does it include data for all of the machines at BTL. It is, however, based on the data from about 150 different machines, and is remarkably consistent over the last year. In general, it reveals that of the screen editor uses, vi is about 2/3, emacs (mostly mine, some Gosling's) is about 1/3, with small amounts of others. This isn't surprising, since vi is more generally supported and has been around longer than any of the various flavors of emacs. Ed is invoked more commonly than everything else put together, but I suspect that a lot of this is due to batch usage of ed, as in shell scripts or various other tools that do a little bit of editing as part of a larger task. The data shows strong pockets of users of one sort of an editor or another. Some machines show all emacs and no vi, some all vi and no emacs, and some neither. In selecting an editor, I agree completely with the perception that editor preference is largly a religion once an editor is used. In my 15 years of computing, I have used the "retype the line by line-number" editor used in BASIC systems, an ed style editor, and Emacs. Each transition was a bitter battle for me to accept a new way of doing things. I was one of the last people in my group to make the transition to Emacs, but became a true believer. My perception is, however, that conversions are relatively rare, and in general people stick with the editor they know and love unless you take it away. (When I left MIT and lost access to Emacs, I re-invented it rather than learn vi or go back to ed.) In making the initial choice, I would think that the following reflects what influences people most: 1) A local Guru. This is why you see pockets of use. One organization will have an emacs Guru that shows each new user how to set up initialization sequences and patiently explains all of the commands to people in their organization. Another will have a vi Guru doing the same. 2) Official Support. Some people don't like experimental things. 3) Personal stye. My perception is that emacs/vi use may correlate with other aspects. People who compose and edit completely on line, without ever producing paper are more likely to favor emacs, while those who write drafts or print paper copy and mark it up for later editing are more likely to favor vi. Some people are more comfortable with the push-a-button-and-watch-what-it-does style of emacs, while some favor a structured command oriented approach to human-machine interaction. I think that the quality of manuals is important in user statisfaction, but not critical in determining how users choose editors. Users are lazy and won't read the manual if the guy down the hall will answer the question for them. -- Warren Montgomery ihnss!warren IH x2494
notes@ucbcad.UUCP (11/03/83)
#R:cca:-601000:ucbesvax:2800001:000:1722 ucbesvax!turner Nov 3 02:33:00 1983 We have dozens of people here (esvax at U.C. Berkeley, a machine shared by EE and ME departments), and emacs comes and goes, usually living in /tmp when it's around, and used by at most 3 people out of a user community of over 100. These people need emacs when their "other" systems are down, or overloaded. The documentation can be found by perusing their personal directories. Everyone else (myself included) uses vi, or ed/ex. I like vi OK, and am still learning a thing or two about it five years after starting to use it. But then I seldom read more than the Vi Quick Reference. I played with Gosling emacs, but got frustrated. The quick ref page helped some, but I lack the patience to plow through the on-line tutorial, which is slow and confusing, or to read the manual, which is only in computer-readable form here. (When it's around.) I am doubly frustrated in that I have a sense of the power of emacs after five years of going out to the edge of what vi can do and thinking, "oh yes: to do that you would need a built-in language interpreter--rats." (Modelessness, on the other hand, has never struck me as an intrinsic virtue.) My experience with learning new systems is that there is an order of magnitude speed-up in having someone close at hand who can tell you how to do things, what you did wrong, and what you should have read before you tried to do something a little tricky. There is, unfortunately, no such person around here. Since I dislike bugging people on other machines for help, or publishing my ignorance on USENET, I'm frozen out until I have a free week or two (hah!) that I might devote to learning the basics of a new editor. --- Michael Turner (ucbvax!ucbesvax.turner)
notes@pur-ee.UUCP (11/06/83)
#R:cca:-601000:ecn-ee:22000001:000:1415 ecn-ee!davy Nov 5 13:13:00 1983 Since we're discussing Emacs usage.... Here at Purdue's EE Dept., we have about three or four copies of Emacs. None of them are "system" copies, they are all in people's personal directories. I have a version of CCA Emacs which is about a year old, and the other people all have (I believe) Gosling's. If I were to estimate how many people used any version of Emacs, I would have to say less than 25 people (out of a user community of roughly 4000). I can site various reasons for this. First, the "system" does not support Emacs, and therefore, news of its existence travels only by word of mouth. Secondly, most of our users are students, and care only about getting their assignment finished. For this, they use the "eed" editor, which is a local editor similar to "ed". Secondly, the recommended screen editor here is "vi". We also have the Rand editor, but it tends to punch the crap out of the system, and its use is therefore discouraged. Finally, the locally available documentation for Emacs is poor. Personally, I find myself alternating between Emacs and vi. I prefer to use Emacs for writing C programs, because it is much more intelligent about indentation, etc. than vi. However, for straight text, I prefer vi. Probably because vi is still somewhat line oriented. Let's face it -- you can argue all day about which editor is best, and still not find out. --Dave Curry pur-ee!davy