Re: Replying to lists (and related queries)

[Brian Stafford <brian stafford uklinux net>]

On Wed, 10 April 09:59 Peter Bloomfield wrote:
> On 2002.04.10 03:44 Brian Stafford wrote:
>> Hi all,
>> 
>> I've been making fairly heavy use of the automatic identity selection 
>> feature of late and it works well for me except in one situation.  Replying 
>> to a list.  Its becoming an irritation because I keep forgetting to 
>> explicitly select the identity before sending messages to the list, leading 
>> to unnecessary bounces, list admin actions etc.
>> 
>> I thought I'd do a patch for this so I found the code that guesses the 
>> identity, but then I got stuck.  Currently Balsa guesses the identity from 
>> the content of the original message's To: header.  Just now, I can't think 
>> of a good way to guess/select an identity for replying to lists.  Any 
>> suggestions?
> 
> I believe you'd need to add a field to the identity, `Use this identity when 
> posting to lists', perhaps with some way to specify one or more lists.

Hmmm.  I like that idea.  Maybe stating "lists" was a red herring though, more 
accurately `Use this identity when replying to these recipients'.  I also like 
the Nutscrape approach of having a default identity per mailbox.  But I see no 
reason that the two are mutually exclusive, they might even complement each 
other quite well, I can see that they will be useful in different situations.

>> Realted thing.  Would it be more sensible to replace the address book 
>> button to the right of the From field with the identity selector button?
> 
> Yes.
> 
>> Allowing the arbitrary selection of From: from the address book appears to 
>> violate RFC 2822.  from section 3.6.2
>> 
>>    In all cases, the "From:" field SHOULD NOT contain any mailbox that
>>    does not belong to the author(s) of the message.
> 
> What does `belong to' mean?

This is somewhat vague, possibly intentionally so.  I take this to mean a 
mailbox the sender of the message is `morally entitled' (my own words) to 
use.  That would include the sender's own mailboxes, e.g. I might send a 
message with Sender: brian@stafford.uklinux.net and From: set to  
brian.stafford@clara.co.uk, I'm morally entitled to the from: address because 
its mine.  Alternatively, RFC 2822 quotes the example of the message author's 
mailbox in From: but which is sent by the author's secretary whose mailbox is 
in Sender:.  Here the secretary has the author's permission to use the From: 
address.  I'm sure there are other examples.  OTOH, I can't see that I have a 
moral entitlement to use PeterBloomfield@MindSpring.com in any message I send 
(unless I've been given permission).  Taking these examples together implies 
that algorithmically enforcing the contents of From: is difficult.  I guess 
its a value judgement on the part of the UA's authors.

>> Another related thing, when using the address book button to pick an 
>> arbitrary From: address, does balsa set the Sender: header from the 
>> identity?  RFC 2822 states that the Sender: header should be set when From: 
>> lists multiple addresses or when the mailbox in From: is not the actual 
>> sender of the message.  Again this is described in RFC 2822 3.6.2
> 
> Balsa doesn't allow multiple `From:' addresses. Presumably it should, but 
> that's another matter.

Indeed.  It would be useful when multiple authors collaborate on a message 
which is sent by one of the authors or another party.  Again this exercises 
the meaning of `belong to'.  Presumably, most of the authors' mailboxes do not 
'belong to' the sender, but permission is granted to the sender to use them.

However what I had in mind here was the more specific case of selecting a 
single alternative From: address from a list of my legitimate mailboxes with 
the expectation that Sender: will be set from the identity automatically (and 
unset when From: names the sender).  Such a list might even differ per 
identity.

I think there are a few issues here that might be addressed by individual 
patches.  I also think that there are different levels of usefulness that can 
be ascribed to each of them.  Right now, I'm really only interested in the 
identity selection thing.

Brian