Hi, If we end up going the K/V route: > On 22 Jul 2017, at 09:53, Viktor Dukhovni <[email protected]> wrote: > > On Fri, Jul 21, 2017 at 12:59:51PM +0100, Jeremy Harris wrote: > >>> I would prefer to have more opinions (implementors in particular) but >> >> There's no way I would add JSON support to exim. > >> On Fri, Jul 21, 2017 at 04:50:46AM -0700, Claus Assmann wrote: >> >> There's no way I would add JSON support to sendmail. > > I'll chime in with no plans to add a JSON dependency to Postfix. > The current KV proposal is fine. I would have avoided the > superfluous ":" characters after the keys: > > mx foo.example.com > > rather than: > > mx: foo.example.com > > but I don't care enough to make this an issue. The authors are > free to drop the unnecessary ":" if they agree.
This has a nice property that I can reuse a generic header field parser. > The K/V would then be a sequence of lines, with one K/V per line, > with each line of the form: > > <KEY><whitespace><VALUE><optional trailing whitespace><EOL> > > <EOL> is whatever goes for EOL in HTTP text/plain content types, > which IIRC is <CRLF>. Yes. > Client implementations MAY be liberal and > also allow <LF>, but server implementations SHOULD only use <CRLF>. No bare LF please, CRLF only. Best Regards, Alexey _______________________________________________ Uta mailing list [email protected] https://www.ietf.org/mailman/listinfo/uta
