From:	MERC::"uunet!CRVAX.SRI.COM!RELAY-INFO-VAX"  9-JAN-1993 13:07:24.65
To:	info-vax@kl.sri.com
CC:	
Subj:	Re: Question: Reading keys with command procedures.

leichter@lrw.com (Jerry Leichter) writes:
> In theory, any undocumented language support routine is subject to change in
> future versions of VMS.  In practice, since the VMS group and the language
> groups try quite hard to ensure upward compatibility, it's essentially
> impossible for them to remove any such routines, or change the semantics of
> calls generated by older compilers.

Well, yes, but the implied support by not breaking existing code only
goes as far as "normal" usage.  The BASIC subroutine I derived my code
from goes:

	sub fred
		x$ = inkey$(0%, wait)
	end sub

and compiles to:

	FRED::	.word ^M<R2,R3,R4,R5,R6,R7,R8,R9.R10,R11,IV,DV>
		movab .-3,R2		; \
		movab $PDATA+4,R0	;  | This is BASIC's init code
		movl R0,R1		;  |
		jsb BAS$INIT_R8		; /
	1$:	movab 1$, -4(FP)	; This is for ERL handing
		pushl #0		; \
		pushl #0		;  |
		pushl #0		;  | I lifted this for my code
		pushaq Z$(R11)		;  |
		calls #4, BAS$INKEY	; /
		movab $PDATA+4,R0	; \
		jsb BAS$END_R8		;  | This is BASIC's exit code
		movl #1, R0		; /
		ret

with data structures to match.  It happens that BAS$INKEY *in* *this* 
*version* runs "standalone"; it's possible however that in the future, data
structures may be set up for BAS$INKEY within BAS$INIT_R8.  Code compiled
with BASIC will be fine, because it contains calls to BAS$INIT_R8, but
direct calls to BAS$INKEY from languages other than BASIC wouldn't. 

BASIC is a particularly nasty example, because of the truly awful code
the compiler generates....  (Nice language, cruddy pessimising compiler
backend.)
 
> Mr. Stokes also says that "all bets are off on Alpha."  For native Alpha

<cringe>  "Don", please.  8-)

> code, that's certainly true, since there is no reason why a native Alpha
> BASIC compiler should generate calls to the same RTL routines as a VAX
> BASIC compiler.  However, to the degree that Alpha VMS continues to support
> the running of VAX images, it will necessarily be subject to the same
> constrains as I pointed out above.

... and it's perhaps more likely that changes as outlined above may have
been made.

--
Don Stokes, ZL2TNM (DS555)                        don@zl2tnm.gen.nz (home)
Network Manager, Computing Services Centre            don@vuw.ac.nz (work)
Victoria University of Wellington, New Zealand              +64-4-495-5052