Re: hmac-md5

[Bruno Haible]

Simon Josefsson wrote:
> > Yes, I try to avoid 'unsigned char *' for that reason'.  ...
>
> Then perhaps hmac-md5 should use 'char*'.

The big downsides of 'char *' are:

  - It is not closer to the meaning of the data than 'void *'.
    The hmac-md5 module considers all bytes as unsigned values. Giving
    a prototypes that presents the bytes are string elements doesn't
    help.

  - It is unportable.
    For example, the hash function for strings in .mo files in glibc
    was taking a 'const char *' as argument and not casting the elements
    to 'unsigned char' - and it led to portability problems.

> I think I'll leave it as 'void*' for now

Agreed.

Bruno