jove vs. vi: performance

[Paul Vixie]

In article <282 at lclark.UUCP> dan at lclark.UUCP (Dan Revel) writes:
|Does anyone know what sort of impact jove has on system performance?

Compared to VI, they're about equal.  Startup time is going to depend mostly
on the size of the executables, which on my Vax (Ultrix 2.x, Jove 4.9, VI 3.7)
are:
	text    data    bss     dec     hex
	115712  5120    125992  246824  3c428   /usr/ucb/vi
	100352  23552   75736   199640  30bd8   /usr/local/bin/jove

Rather than Jove, you're probably worried about emacs-type editors in
general, since their text and/or data frequently gets to 600K or 1MB --
then it starts running and eats up lots and lots more.  Emacs editors,
in general, are written in Lisp and contain a byte-code interpreter.
This makes for an amazingly programmable editor, but it also eats up
all available CPU cycles and memory.

Jove does not suffer from this problem, as it is written in C and has
no e-lisp or m-lisp or any-kind-of-lisp extension language.

Let's not, everybody, get into any kind of a discussion of which editor
is better.  If someone wants to dispute what I've said about resource
usage, fine.  But if I see another "vi vs. emacs" subject line this week
I may be forced to eat my own liver.

In article <24356 at bu-cs.BU.EDU> madd at bu-it.bu.edu (Jim Frost) writes:
# You might look into microemacs, which is more portable than jove, more
# functional, and is still quite small and quick.  That should also have
# little effect on system performance.

I havn't poked into uEmacs for a while, but when last I did I found it
to be flinkier on the whole than Jove was.  Jove is a nightmare internally,
but if you can make it work (lucky me, I have a Vax running BSD :-)) it
works quite well.  uEmacs was my favorite editor when I worked on MSDOS,
but Jove works even there these days.

If anyone wants to talk about jove vs. uEmacs, let's go to comp.emacs or
to comp.editors, but please, please, please, don't do it in comp.unix.q.

# BTW, 15 people on a 11/750 is going to be very, very slow no matter
# what editor you're using.

If I were going to put 15 people on a machine and have them all edit files
simultaneously, I'd choose a 750 before a MicroVAX-II.  The II may run rings
around a 750 in single-user benchmarks, but once a lot of users get on there,
it degrades much more steeply.  Whether this is because of the IO bus or
standard-configuration disk subsystems or just what, I don't know.
-- 
Paul Vixie
Digital Equipment Corporation	Work:  vixie at dec.com	Play:  paul at vixie.UUCP
Western Research Laboratory	 uunet!decwrl!vixie	   uunet!vixie!paul
Palo Alto, California, USA	  +1 415 853 6600	   +1 415 864 7013