Re: Clarifying the qualities we desire the DIGEST-MD5 replacement to have

[Alexey Melnikov <alexey.melnikov@xxxxxxxxx>]

Simon Josefsson wrote:

> Very useful summary.
+1.

> I agree with all those requirements, except that I have some reservation
> about using a textual protocol.  My experience with DIGEST-MD5 has been
> that the free format created more problems than it solved.  There are
> still interop problems with quoted text.  I recall interop problems
> caused by the specification permitting empty comma sequences like ", ,".  
Indeed.

> If it is a textual protocol, I would want it to only permit ONE way to
> say the same thing.
>
> Reordering parameters so that 'foo=bar,baz=apa' is
> equivalent to 'baz=apa,foo=bar' is bad.  Permitting whitespace so that
> 'foo=bar, baz=apa' is also equivalent is bad.  Using quoting, so that
> 'foo="bar",baz=apa' is also equivalent is bad.  Allowing unrecognized
> parameters to be skipped is bad.  And so on.  
Agreed.