Date: Mon, 18 Nov 1996 12:36:23 -0800 (PST) From: Chris Newman <Chris.Newman@innosoft.com> Subject: Re: IMAP URLs In-Reply-To: <"josef.ifi..985:18.10.96.09.23.48"@ifi.unizh.ch> To: Martin J Duerst <email@example.com> Cc: URI list <firstname.lastname@example.org> Message-Id: <Pine.SOL.3.95.961118122912.2248Hemail@example.com> On Mon, 18 Nov 1996, Martin J Duerst wrote: > The second problem is somewhat more basic: If you want to change > a list of parameters, how do you indicate the absence of a parameter? > One possibility would be to indicate the parameter again, with an > empty value, but in some cases, an empty parameter value and the > complete absence of the parameter will not mean the same. That is a problem. > Therefore, it looks very much like the specification for not > partially inheriting parameters as in RFC 1808 makes a lot of > sence, and is definitely not a bug, even if in some cases, > it may not exactly be optimal. Where the problem comes with IMAP URLs is the ;AUTH= parameter used to specify an authentication mechanism. This would almost always need to be copied to relative URLs, whereas the ;UIDVALIDITY= and the ;UID= parameters would be much less likely to be copied to a relative URL. Now if there were some way for me to put the AUTH= in with the <login> part of the URL, it would be a lot cleaner, but I can't think of a good syntax. I'd also say that as we move away from cleartext passwords, it may become more important to specify authentication mechanisms in URLs in general.