Subject: CRLF fun stuff again...
From: John R. Dunning (jrd@jrd.org)
Date: Tue Feb 13 2001 - 00:07:23 EST
[Getting to be off-topic; sorry...]
From: "Chris Garrigues" <cwg-dated-976de3b93a5c80b6@deepeddy.com>
Date: Mon, 12 Feb 2001 17:03:51 -0600
[...]
Make would be next to useless w/o extensions.
Correct. It's extremely rare any more to find a makefile which
doesn't rely on pattern rules, eg rules driven by semantic knowledge
of filename extensions.
[...]
The use of extensions (and a lot of other things)
in MSDOS was based on the same in CP/M and CP/M got most of those same things
from RT-11 (or was it RS/X?).
RT had variants with no filesystem, as well as extremely simple
filesystems (not much more than a disk driver), so it's somewhat
arguable whether extensions were a feature there. RS/X (often
pronounced R-sucks :-) made heavy use of filename extensions as far
back as the early '70s or so.
Most of the various competing minicomputer OSes also used extensions in a
similar way.
Right. TOPS-10/20 in particular was an early player in that game;
that one dates back (I think) to the late 60s; you can argue that it's
not a competitor, though the groups at DEC would likely have
disagreed. I remember hacking on other late-60s early-70s systems by
long-forgotten outfits like Interdata and Harris Microsystems; they
all had filename extensions that meant something.
*nix was notable in its early days for being the maverick, and *not*
having any kind of convention for carrying file type information out
of band from the data. Since it (later) completely overwhelmed the
industry, it's currently fashionable to forget that there were earlier
systems which did a much more comprehensive and consistent job of
dealing with the issue.
I've been doing this *way* too long :-{
This archive was generated by hypermail 2b28 : Sun Oct 14 2001 - 03:04:32 EDT