Re: Elaboration_check For Instantiations

[dewar@merv.cs.nyu.edu (Robert Dewar)]

Charles Sampson says

<<     You are right, the GNAT implementation is entirely correct.  I
must have mislead you when I said "technically correct".  Technically
correct is correct, no doubt about it.  All I meant was that GNAT is not
following the programmer's wishes.  If Suppress(Elaboration_check) is
written, then the programmer doesn't want the check made; if the check
is not made then Program_error can't be raised.  The programmer's wisdom
in making such a request is another discussion.

                                Charlie>>


No, that's an incorrect viewpoint. The programmer is NOT saying that
they do not WANT the check, they are saying they do not NEED the check.
(because they know that the checked condition cannot occur). That is
a HUGE difference.

pragma Suppress (x)

does NOT mean "please suppress all checks for x", it means "I know my
program has no instances of violations of check x, so feel free to
omit checks for x if this will speed up the code."

Note that if a program with Suppress fails a check, then it is erroneous,
so a programmer cannot be expressing a wish about how the erroneous program
is to be treated.

A programmer who writes pragma Suppress not understanding this is not
"unwise", just ignorant of the meaning of Suppress in Ada, a common
failing of understanding.

GNAT is *exactly* following the programmer's wishes, but only if the 
programmer knows Ada! If the programmer is writing things down in
Ada thinking they mean something different from how they are defined,
then yes indeed, GNAT hopefully will *not* be following the programmer's
confusion.

I know, the name Suppress is somewhat inherently confusing, and undoubtedly
leads many programmers into the same confusion as Charlie has. Hard to know
what a better name might have been ... something like

Check_Not_Required (xxx)

??

Robert Dewar
Ada Core Technologies