[tlbuild] lcdf-typetools-2.78 -- is hashcode() unsigned, uint32_t, or size_t

[Peter Breitenlohner]

On Tue, 19 May 2009, Karl Berry wrote:

>    Hi Peter, Did you ever get a response from Eddie?
>
> No, he hasn't replied.  Although I didn't say so (sorry), I wrote to you
> and Angelo after Peter and I had discussed it.
>
>    >> -    static unsigned hashcode(const char *begin, const char *end);
>    >> +    static uint32_t hashcode(const char *begin, const char *end);
>
> The point is that the prototypes use one of unsigned/uint32_t and the
> definitions use the other?  As Peter wrote originally, it would be good
> to explain the actual problem to Eddie.  On the face of it, it's not
> clear that why uint32_t is preferable.

My main problem with the change was actually this:

using uint32_t without *explictly* including the header(s) supposed to
define it make this a but fragile, depending on the fact that these are
included implicitly.

Thus I wanted Eddie Kohlers opinion, how that should be handles.

Regards
Peter