🚀 go-pugleaf

gavron@spades.aces.com (Ehud Gavron 602-570-2000 x. 2546) writes:

>In article <1992Dec13.020257.1@tango.cchs.su.oz.au>, brennan@tango.cchs.su.oz.au (Luke Brennan) writes...
>#In article <12DEC199200264446@spades.aces.com>, gavron@spades.aces.com (Ehud Gavron 602-570-2000 x. 2546) writes:
>#> In article <Bz3KEC.Es2@wm.estec.esa.nl>, hevans@estcs1.dnet.estec.esa.nl writes...
>#> # 
>#> #Can anyone out there tell me if it is possible to write my own lexical
>#> #functions, and how?
>#> 	The answer to your implied question is NO, you cannot normally
>#> 	write your own lexical functions into DCL.
>#	Why precisely? No hooks? Yes, but undocumented? Totally impossible?
>#	'you cannot normally'.. okay - so what about NOT normally? :-)
>The other lines of the post which listed how you could, and why you NORMALLY 
>can't were quite clear.
>#	Isn't this the kind of thing an internals guru would *eat* ? !!
>#        Seems like an incredible oversight for VMS - I recall even PRIMOS
>#	had that kind of thing standard in the RTL..
>There is a big difference between running an image, and adding your own 
>iser functions to a privileged interpreter.  That is one Pandora's box
>that Digital has left unopened.  It's too bad, of course.  It would be
>nice if it were there, but there's nothing an external image can do that 
>a lexical could do much better.
>#	Just asking.. Luke.
>I'm sure you realize that that question (as well as why there is no DCL
>compiler) has been asked many times.  The question is not stupid, nor
>is the answer: it opens up too many cans of worms to allow user modification
>(including putting in the appropriate hooks) to the command interpreter 
>which everything in VMS depends on, from bootstrap sysinit phase and on.
>Ehud

Ehud,

	It is hard to argue with somebody who is right, as you are in this
case.  SO let's get down to cases and ask what kind of lexical functions
people actually want.
	I, for one, would like a complete time arithmetic package.  Yes,
I know I can run a program, use the $ASCTIM and $BINTIM system services and
the LIB$ADDX and LIB$SUBX calls to do the time arithmetic.  But, doing all
that takes an image activation, plus, I gotta get the symbols out of DCL and
put the answer back into a symbol in DCL... a mess.
	I would like a silo of the last few (20? 30?) error messages, including
the 32 bit error codes.
	I would like a lexical function which returns the CPU type (Alpha, or
VAX).

	Jeff Silverman, Boeing

>--
>Ehud Gavron        (EG76)     
>gavron@vesta.sunquest.com