[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Extraneous file reads
- To: Kent M. Pitman <kmp at MIT-MC>
- Subject: Extraneous file reads
- From: MOON at SCRC-TENEX
- Date: Mon ,5 Apr 82 17:58:00 EDT
- Cc: BUG-ZWEI at SCRC-TENEX
Date: 4 April 1982 21:50-EST
From: Kent M. Pitman <kmp at MIT-MC>
Subject: Extraneous file reads
To: BUG-ZWEI at MIT-MC
When I have a file FOO.BAR#> and I explicitly write it to FOO.BAR#100 (which
is higher than any existing FOO.BAR), then a Find File occurs (eg, via Tags
Search), Zwei finds out it needs to use FOO.BAR#100, so reads it from disk
rather than using it out of my buffer.
Actually Zwei finds out it needs to use FOO.BAR#>, which is not in a
buffer; the buffer you have is explicitly for version 100 and the greatest
version might be bigger than 100. c-X c-W changes the file name associated
with the buffer. You can use m-x Set Visited File Name to change the file
of the buffer back to > if that's what you want (i.e. you just wanted to
crank up the version number, not to keep that version explicitly in a
buffer.)
Additionally, M-X Compile File always prompts for a buffer name to compile,
yet last time I used it -- many versions ago -- it wanted to always read off
of disk even when there was an unmodified copy in core. If it's going to use
the copy off of disk, it might as well prompt for a buffername instead of a
filename. I stopped using M-X Compile File as a result of this; now I just
use QC-FILE or QC-FILE-LOAD from lisp listeners because they seem to do more
natural things. I wish these were Zwei commands...
m-x Compile File is precisely QC-FILE. In recent Symbolics software the misleading
prompt for m-x Compile File was fixed and m-x Load File was added, which is
a Zwei command version of QC-FILE-LOAD (it only offers to compile if the source
has changed).
Accessing files from remote file systems is very slow and cleaning up little
points like this save a lot of time and hassle ...
It sounds like you're suggesting that when the machine goes to open a file
for input, if that file has been read into an editor buffer it offer to
read from the editor buffer instead? I guess that could be done although
probably it is only significantly faster when the file is on the AI
machine.