Charconst valuation and short-circuiting

Charconst valuation and short-circuiting

[Neil Booth]

Should GCC not be attempting to value the character constant in
the following code (and therefore not give a hard error about
a too-long charconst?)

  if (0 && 'abcdefghijk')
   ...

Neil.

Re: Charconst valuation and short-circuiting

[Alexandre Oliva]

On May  6, 2002, Bonzini <bonzini@gnu.org> wrote:

>> Should GCC not be attempting to value the character constant in
>> the following code (and therefore not give a hard error about
>> a too-long charconst?)

>> 
>> if (0 && 'abcdefghijk')

> I don't think so.

Agreed.  IMO, it's not too different from:

  if (0 && undefined_variable)

which is surely expected to trigger an error.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer

Re: Charconst valuation and short-circuiting

[Bonzini]

> Should GCC not be attempting to value the character constant in
> the following code (and therefore not give a hard error about
> a too-long charconst?)

> 
> if (0 && 'abcdefghijk')

I don't think so.  It is not unlikely to why you use if's instead
of #ifdef's so that dead code is eliminated but still flags errors.
If you wanted no errors, you'd use a preprocessor condition (I'm assuming
that 0 is valuable at compilation time, and hence at preprocessing time,
maybe with some help from autoconf a la SIZEOF_LONG).

Paolo