To: drums@cs.utk.edu
Subject: Re: Case sensitivity
References: <7JbPI8Smw-B@khms.westfalen.de>
From: Russ Allbery <rra@stanford.edu>
In-Reply-To: kaih@khms.westfalen.de's message of "26 Jun 1999 08:52:00 +0200"
Date: 26 Jun 1999 03:06:39 -0700
Message-ID: <yllnd7qnrk.fsf@windlord.stanford.edu>
Lines: 27

Kai Henningsen <kaih@khms.westfalen.de> writes:

> A recent exchange in USEFOR brought up that 822bis doesn't contain any
> clue about case sensitivity of addresses and message ids. For addresses,
> it points to 821bis which defines the LHS to be case sensitive, and the
> RHS to be a domain; I expect the DNS RFCs somewhere explain that that
> one is case insensitive.

> However, it is pretty far from obvious to apply all this to message ids.

> I believe something must be said about this, or we'll get
> interoperability problems.

And to add some background to this, INN has for years been treating the
LHS of message IDs as case sensitive and the RHS as not case sensitive,
due to the requirements of RFC 822 (which defined a message ID in terms of
an addr-spec, and therefore sort of implicitly picked up the case
sensitivity from e-mail addresses).  Looking at the code, it currently
even special-cases "postmaster", but doesn't fully handle quoting.

For the purposes of news, it would be preferrable to just say that message
IDs are case-sensitive, but that compliant systems SHOULD NOT generate
message IDs which differ only in case on the RHS.  (Differing only in case
on the LHS is fine, ignoring the edge case of "postmaster".)

-- 
Russ Allbery (rra@stanford.edu)         <URL:http://www.eyrie.org/~eagle/>