Strangeness in ASRT

Strangeness in ASRT

[Richard Stallman]

This seems to have a bug which breaks compilation.

    /* The IDX==IDX tries to detect when the macro argument is side-effecting.  */
    #define ASET(ARRAY, IDX, VAL)	\
      (eassert ((IDX) == (IDX)),				\
       eassert ((IDX) >= 0 && (IDX) < ASIZE (ARRAY)),	\
       ASLOT ((ARRAY), (IDX)) = (VAL))

Changing ASLOT to AREF makes it compile.  (Bootstrapping still crashes
as before.)

What is the purpose of these tests?  The expansion of AREF
substitutes IDX only once, so there is no problem if it has
a side effect.

Re: Strangeness in ASRT

[Stefan Monnier]

> This seems to have a bug which breaks compilation.
>     /* The IDX==IDX tries to detect when the macro argument is side-effecting.  */
>     #define ASET(ARRAY, IDX, VAL)	\
>       (eassert ((IDX) == (IDX)),				\
>        eassert ((IDX) >= 0 && (IDX) < ASIZE (ARRAY)),	\
>        ASLOT ((ARRAY), (IDX)) = (VAL))

> Changing ASLOT to AREF makes it compile.  (Bootstrapping still crashes
> as before.)

Oh, yes, it should be AREF, not ASLOT.  Sorry for the mixup.

> What is the purpose of these tests?  The expansion of AREF
> substitutes IDX only once, so there is no problem if it has
> a side effect.

If ENABLE_CHECKING is true, then eassert ((IDX) >= 0 && (IDX) < ASIZE
(ARRAY)) will cause IDX to be eval'd several times, so you'll get errors
when you set ENABLE_CHECKING and an argument has the form "index++".


        Stefan