[Python-Dev] gcc barfs on recent stringobject changes...

[Tim Peters]

[Mark Favas]
> Changes in the last few hours (hi Tim!)

Hi Mark!  Sorry about that!

> to stringobject compile (I'd guess) on MS

You guess right -- and under two flavors of Windows <wink>.

> (and on Compaq's Tru64 compiler),

Figures.

> but produce the following with gcc on Solaris and FreeBSD:
>
> gcc -c -g -O2 -Wall -Wstrict-prototypes -I. -I./Include
> -DHAVE_CONFIG_H  -o Objects/stringobject.o Objects/stringobject.c
> Objects/stringobject.c: In function `PyString_FromStringAndSize':
> Objects/stringobject.c:76: invalid lvalue in unary `&'
> Objects/stringobject.c:80: invalid lvalue in unary `&'
> Objects/stringobject.c: In function `PyString_FromString':
> Objects/stringobject.c:130: invalid lvalue in unary `&'
> Objects/stringobject.c:134: invalid lvalue in unary `&'
> *** Error code 1

Fair enough:  I tried to use a cast as an lvalue in those 4 places, all of
the form:

    		PyString_InternInPlace(&(PyObject *)op);

where op is declared PyStringObject*.  Strictly speaking, that ain't legal,
but changing it to:

		PyObject *t = (PyObject *)op;
    		PyString_InternInPlace(&t);

is.  You may wonder WTF the difference is.  That's easy:  the rewrite doesn't
use a cast expression as an lvalue <wink>.

sensible-or-not-it's-checked-in-so-please-try-again-ly y'rs  - tim