Re: Mismatch about which version of C is ok to require (K&R woes, again)

[Stefano Lattarini]

On 01/01/2013 11:29 PM, Karl Berry wrote:
> Hi Stefano,
> 
>         To support pre-standard C, instead of writing function definitions
>         in standard prototype form, [SNIP]
>     and then again, near the end:
>         In order to support pre-standard compilers that do not recognize
>         prototypes, you may want to use a preprocessor macro like [SNIP]
> 
> Sure, both of those are only talking about *if* you want to support
> pre-standard C.
> 
>     both of which seems redundant, if not in contradiction, with the fact
>     that it's now OK to require C89.
> 
> It is neither redundant nor a contradiction, as far as I can see.
>
Ah OK, so the standards don't say "don't bother supporting K&R, ever",
but only "it's perfectly OK not to support K&R; if you really want to
do so, follow this advice ...".  If this is the case, I can see there
are indeed no contradiction (maybe the wording can be clarified a
little to avoid confusion like mine, but that is low priority and
maybe not even worth the effort).

>     IMHO this references to no-longer-supported K&R C (and other similar
>     references, if any) should just be removed.
> 
> Last I knew, rms doesn't want to remove K&R C references yet, because
> it's not true that it is "no longer supported" (in his mind).
>
Bummer.

> What's true is that GNU packages are not *required* to support K&R.
> (And haven't been for a long time.)
>
OK, got it.

Thanks, and sorry for the noise,
  Stefano