Alan@lcs.mit.EDU (Alan Bawden) writes:
> It's possible that a cool notation solves this problem neatly. But I think
> it much -more- likely that a procedural toolkit can solve it neatly.
Um, so write and release it?
There are a couple of advantages to Olin's current scheme:
#1: it exists (I would have liked to have had it
available earlier this evening, actually)
#2: it seems to straightforwardaly make it easier
for me to get in touch with the bare metal of
UNIXy things normally done with ugly shell scripts
in inferior scripting languages
#3: it does not seem to be hopelessly flawed,
although there have been some suggestions for improvement
which do make sense
I dunno about what Olin thinks, but I think that in the
spirit of UNIX it should go out the door for use and if
this spurs on the development of something better, great.
If on the other hand it proves sufficient (even if not
optimal) and therefore nobody builds a more successful
better mousetrap, well, then that's just how UNIX has been
for a long time. I use successful to mean "is more widely
Of course, this is heresy to anyone who religiously wants
scsh to be a more successful better mousetrap than other
UNIX shells and even whole sets of UNIX tools, but then
I'm of the opinion that reimplementing everything that's
UNIX from the ground up in a decent language is a much
more fun approach. scsh, to me, therefore, is merely a
useful temporary device for making UNIX much more bearable
for a range of things right now. Spending so much time
"perfecting" designs for scsh-y things that they never get
released -- from this point of view -- is bad.