On Tuesday 15 August 2006 13:16, jdow wrote:
Remember that carrier pigeons have a higher RFC number, today, than your precious 2369.
If I want to send internet packets on a pigeon, I'm going to use that approach rather than inventing a new one of my own, for sure. Why go to all the trouble of reinventing the wheel (or in this case, the bird)? Similarly, since there are several published and well known descriptions of how to formulate internet messages, including details of extra header information
Default behavior has been to use the Reply-to header if it is present falling back on the From header.
For the "Reply-to-Author" pattern, yes. "Reply-to-List" should clearly do something else, and a generic "Reply" function should do what the user configures it to do.
Making a change in the default behavior is bad. And making the default behavior different than users expect is bad. Hence presuming that the List- headers in 2369 mean they are to be used in place of Reply-to is a little ingenuous to say the least.
Disingenuous? Nowhere did I suggest that List-Post should be used to _override_ Reply-To, only that using Reply-To _instead_ of List-Post to transmit information inserted by the list server is inappropriate. If a list message arrives for me which contains a Reply-To header, and I press reply, then my reply is addressed to the Reply-To address, as you would expect. If Reply-To is absent, it is addressed to the List-Post address, if it came from a list which correctly sets the RFC2369 list headers (I could turn this off, but it is exactly the ideal list-reply behaviour users of nonconformant clients perform workarounds to achieve). Otherwise, it is addressed to the From address.
So, in the situation that the list conforms with RFC2822 and RFC2369, and the client also conforms with RFC2822 and RFC2369, everything works exactly as you suggest (my client is configured (by default) so that when I press "R", meaning "Reply", to a list message, the reply is addressed to the list). The list sets List-Post, and the client is configured to send replies to list messages to the list.
Incidentally, the extra action you mention is simply an alternative action - "Reply to Author" instead of "Reply". That's exactly the action which is broken if the list sets the Reply-To header.
No - that violates the principle of "least surprise." It is bad UI design.
You've lost me here - which part violates that principle? (or, to put it another way, what is the "least surprising" behaviour for a "Reply-To-Author" command, and that for a generic "Reply" command, variously in the presence or absence of Reply-To and the various list headers, in particular List-Post?) I have various options when I reply to any message, all of which do what they claim to do (unless a header which is supposed to contain information about the author has been misused, and in fact contains information about a list server, for example). My standard "Reply" command honours Reply-To, if present, and also uses List-Post to facilitate sending replies to list messages to the list in the absence of a Reply-To header. If I wish to perform one of the less common paths, this is easy too - "Reply-to-List" allows me to send to the list address even if Reply-To was set, "Reply-to-Author" allows me to email the author directly at his Reply-To (if present) or From address, and "Reply-to-All" (probably From, Reply-To, To, can't remember since I rarely use this one) are straightforwardly available. It does this without any even vaguely controversial use of any message header, by using the published specifications of a proposed standard way to do this, which seem to work very well.
Lucky the list doesn't alter the Reply-To header, or I wouldn't have been able to divert this conversation to the off-topic list (to which I do not subscribe ;)
Didn't work. I rewrite the headers on the way in.
So you effectively always ignore the Reply-To header, on any email from the list? Well, that's your loss I suppose.