From mit-haydn!leary Sun Sep 11 23:35:36 1983
To: mit-ems!bv
Subject: of sorts
Cc: mit-ems!leary
From: (Dan Leary at MIT-HAYDN)

The following isn't necessarily exhaustive, but it does represent all the
(significant) differences that I've been able to uncover between the two
versions of "sort" in the last few days (except for new features and
improvements in error checking):
	The C version of sort:
- puts a blank space immediately after the opcode.
	(Apparently to no detrimental effect; big score.srt files are just
	a little bigger.)
- doesn't append a final "e" when missing from the source score.
	(Orch requires one.)
- doesn't automatically carry properly (though it tries) when given
	a bare opcode.	(Results in pairs of identical i-cards.)
- writes entire "s" and "e" lines verbatim; comments and all.
	(Orch tries to interpret characters following any "s"
	statement as opcodes.)
	((The manual says: "s anything" and "e anything"
	   -orch only ignores floats.))
- doesn't flag all cases of ramps without a beginning value
	("illegal backward reference") as errors.
	When immediately preceeded by a function card, its corresponding
	p-field is substituted in the beginning of the i-card's ramp.
- doesn't flag all "unresolved" pp's in the same way.

	I haven't really had a chance to get into the code yet, so I can't
really point to where these things occur, but I'm sure you have some ideas
where they happen.  If you'd like, I can probably track down and fix the
ones that need fixing.	Also, I haven't had a chance to look at extract yet,
but will as soon as I can.


