[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: complexity and #$#text-line

I agree with Dave on this one. Lets punt it. 

   Date: Wed, 23 Jul 1997 12:07:07 -0400 (EDT)
   From: Dave Kormann <davek@research.att.com>
   References: <199707231553.LAA28275@arran.research.att.com>
   Content-Type: text
   Content-Length: 1499

    > May I urge people to really really keep it simple, very very
    > simple? This needs to get out as soon as possible, and I very
    > much liked the original spirit of putting in as little as
    > possible. Lets stick to that.

   i think in light of this comment i should clarify my feelings on the
   text-line idea (which seems an area where we're on the verge of adding
   more syntactic complexity), in order:

   my first preference: we don't include it, and we wait to have a
   protocol that does it.  this really seems like an area where a
   separate protocol would work just fine.

   second preference: we include it as a protocol on the level of
   cords: recommended but not required.

   third preference: it goes in the spec proper.

   i guess my objections to this are twofold: first, it seems like
   optimizing the spec to permit one (small) possible parser
   improvement.  second, inband text is ALREADY legal on an MCP
   connection, and your parser is perfectly free to treat such text as a
   special case of an mcp message, just without a '#$#' at the
   beginning.  heck, if you really want to have everything look like mcp,
   you can prefix every incoming line that doesn't have a '#$#' with
   '#$#text-line auth-key text:' before it gets to the parser.  nothing
   about mcp stops you from doing this.

   i'm not at all dead set against this, and if there's general feeling
   it's a good idea, i'll be happy to add it, so don't take this rant as
   anything more than me trying to make clear why i have a problem with