On Aug 6, 2012, at 12:03, Martin Truebner wrote:
>
>>> Dsect Length Id Defn
> ABC 00000010 FFFFFFFF 1
>
> and
>
>>> L'ABC gives me 1 (one) rather than what I want, 16 (10h).
>
> I was as surprised as you were- only the moment was different - I
> looked at length of DSECT in an ADATA-exit and found that HLASM does
> know the length of a DSECT (and so shows the listing in the DSECT
> XREF).
>
> Question- would it really change anything (to unusability), if HLASM
> would return the proper length and not 1?
>
o It would be more useful, but it would surely break some
existing code.
o Some programmers resort to:
DUMMY DSECT
... , stuff
NAME EQU DUMMY,*-DUMMY
o But are length attributes even today allowed to be >256?
(See first bullet.)
o Remember, all this art originated in assemblers that didn't
do any lookahead. Even today, will HLASM do any lookahead to
define a length attribute? Much could be solved by defining
data areas before executable code, but many programmers don't
like that.
On Aug 6, 2012, at 12:48, John Ehrman wrote:
> Worth considering; added to the HLASM "Wish List".
o It would need to be controlled by an option, either global
or on the DSECT statement in order not to break existing
code.
-- gil
