On Sun, 28 Oct 2012 14:10:00 -0500, John Gilmore wrote:
>
>[...] and of course I agree.
>
Worse yet, Adrian Stern agreed recently with something I said on
another list. What am I doing wrong?
>His question about the function of the colon in C followed by the
>assertion that he is familiar with it use only in
>
>| conditional expressions such as "b ? x : y"
>
>The chief function of the colon in C is to delimit statement labels,
"Chief"? Statistically, not in my code.
>as in that immortal TCPL 2 construction
>
>if (a[i] == b[j]) goto found ;
>. . .
>found: . . .
>
I'd probably do this with "break".
>Mr Gilmartin is of course free to deprecate GOTOs ritualistically; he
>has indeed done so here frequently; he should know their syntax.
>
Correction accepted. "Ritualistically"? I'll readily aver that a serious
deficiency of Rexx is lack of a construct such as "longjump" to exit
a nest of function calls. But equally readily that a serious deficiency
of HLASM conditional assembly is lack of any better construct than
AGO. (Could one with macros synthesize AELSE, AENDIF, ALOOP,
etc., and conceal the AGOs?)
I'm delighted that JCL eschewed GOTO when it added IF, ELSE,
and ENDIF. (But I understand the technical reasons that it could
hardly do otherwise, nor provide any iterative construct. Although
I generate much of my JCL with shell scripts in which I freely
employ iteration.) (As delighted as I can be with JCL, that is.)
John M. cited (I didn't trace the thread back very far):
__asm("x DC F'0' ":"XL:DS:4"(x));
... no hint of "goto". But it might be a macro invocation, and a macro
is free to expand as it chooses.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
