On Thu, 25 Oct 2018 14:58:40 -0400, Charles Mills wrote:

>I think I respectfully disagree. Sometimes particular combinations of options 
>yield unspecified behavior. Sure, computers are generally deterministic but 
>the results might be dependent on region size or some other "irrelevant" 
>variable.
> 
Mostly agree.  But once I was furious with a compiler vendor when my program
got fixed-point overflow when optimized and operated as I intended when I
compiled in debug mode.  Vendor said WAD; the construct I depended on was
documented as unpredictable.  Grrr.  Any errors detected in production mode
should be reported in debug mode, especially if the unspecified behavior is
deterministic.

(In debug mode, the compiler generated STH; LH of a halfword variable not
otherwise referenced, truncating an invalid halfword value.  Even before that,
I had felt that by design STH should program check on such a condition.)

-- gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN

Reply via email to