[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

news from the west

Date: Mon 25 Feb 85 17:39:00-PST
From: Rob Austein <G.SRA@SU-SCORE.ARPA>
Subject: found on score
To: cent@MIT-MC.ARPA
thought you might be interested in this.... --r
I have a new version of EMACS that I would like to install on the various
SU TOPS20 systems as soon as possible.  However, this new EMACS contains a
lot of changes from the version that is currently running, so I would like
some of the more daring people out there to run the new EMACS and see if
they can find any bugs before I subject the general user community to this

To run the new experimental version of EMACS, use the following commands
(or include them in your LOGIN.CMD file):


Here is a brief outline on what is new in this version of EMACS.

Based on VERSION 163 EMACS (SU is currently using 162).  According to MIT,
V163 contains many undocumented changes from V162, that may some day be
documented... Sigh.  From this point, even more has been done, enough that
we are calling the current version V165.

Self-updating marks.  If you delete text that precedes a mark, the marks
	will all be updated so that they continue to point at the same

new terminal types.
	AVT, SUN, FR100, FR200, ANSI, AVTX, etc.

Performance improvements:
The performance improvements are the biggest change in this version of
EMACS.  Although many of these improvements will have little effect on
apparent user-level performance (eg they will not result in faster screen
updating), they are expected to significantly lessen the impact of EMACS on
the system as a whole - for a given number of EMACS users, the system load
average should be less...  Here is a breakdown of the areas that have been

  o  TEXTI
	Studies of EMACS use have shown that simple text entry accounts for
	most of the typein directed at EMACS (and editors in general).  The
	TOPS20 monitor and EMACS have both be changed to allow an entire
	line of text to be entered before it becomes necessary to wake up
	the users process and execute actual EMACS code.  The largest
	benefit from this is probably that echoing will occur at interrupt
	level and be essentially instantaneous regardless of system load
	(not including network delays, of course).

  o  Insert mode.
	EMACS has always operated under the assumption that terminals have
	an "insert character" operation, whereas most recent terminals have
	an "insert mode" instead.  By utilizing this feature, it is
	possible to reduce the number of characters that have to be output
	for each character inserted.
  o   SOUT improvements.
	Another study of EMACS performance pointed out that sending
	characters to a terminal is sort of an Achilles heel for TOPS20.
	Every character output was carefully looked at to figure out just
	how it should be output.  TOPS20 has been changed to do these check
	on a per-string basis (rather than per-character), which
	approximately halves the cpu time required to output a string for
	certain cases.  EMACS has be updated do use these forms of output
	more often.