Making it a new method FORCES the protocol to be different and hence
detectable by firewalls (the merit of this I am not discussing).
The point was that we used HTTP so that it would go through proxies
(definitely a good thing) as much as going through firewalls (debatable).
Many proxies only carry port 80.
> -----Original Message-----
> From: Keith Moore [SMTP:moore@cs.utk.edu]
> Sent: Friday, May 29, 1998 3:49 PM
> To: Paul Moore
> Cc: 'Keith Moore'; ipp@pwg.org; moore@cs.utk.edu
> Subject: Re: IPP> review of IPP documents
>
> > Aha the good old POST vs PRINT issue.
> >
> > REQUIRING a different port number would be wrong. We dont preclude this
> > however (we have tested our implementations with non port 80 IPP
> agents).
>
> I disagree. IPP is a different service than vanilla HTTP; there's
> nothing wrong with having separate default ports for each,
> any more than having different default ports for telnet and whois.
>
> (Nobody's required to prevent the use of port 80; it's just that
> IPP needs its own default port assigned to it, and the IPP URI
> needs to default to that port)
>
> I think this is cleaner overall than using a new HTTP method on port 80.
>
> Keith