T in reality

[Barry Shein <BARRYS%MIT-OZ@MIT-MC.ARPA>]

	I hate to inject a strong dose of reality into this
	discussion but I feel obligated to share my recent
	experience with T with this community:

	We recently (Jan) went from 4.1 to 4.2bsd. I gen'd
	in the 4.1 compatibility mode. One of the few programs
	that would not run was T. I sat down, looked at the
	sources and prepared to maybe make a few small fixes
	and recompile. It is not 'written in T' nearly to the
	degree claimed, tons of it is written in LAP
	(ie. assembler with parens around it.)

	Worse, we couldn't recompile it on a 4.1 system (Harvard's
	VAX.) Worse, when we called YALE they seemed disinterested.

	For example, the makefile provided refers to TWENEX files
	and (mostly) unix stuff.

	Result, a graduate student with lots of beached code.
	He finally got it to work by some obscure patch but
	I would think twice before relying on T, the software
	engineering behind Franz make it MUCH more reasonable
	to use in my opinion. The comparisons pale against this
	kind of lossage.

	PostScript: If, for some reason, I was working with some
	totally unrepresentative version of T, my apologies. It
	is possible (I relied on the grad student to get me the
	necessary code) but I doubt that is the case.