Re: mailbox format(s)

On 26-feb-04, at 14:22, Guðbjörn S. Hreinsson wrote:

>> So? We're not saying "you're not allowed to use any other format than
>> this", just "if you want to interoperate with other implementations,
>> this is the format you should use for that".

> I think SHOULD standards SHOULD be refrained from as much as
> possible, these items have in the past generated incompatibilities.

You are taking a single word out of context. Incompatibilities come up 
when standards are unclear. Whether you "should" use a standard or 
"must" use it has no impact on the clarity of the standard. If you feel 
the word "should" must not appear in standards documents I have no 
problem with that.

>>> differences in line endings, bit- and byte-ordering, etc. ad 
>>> infinitum

>> This stuff should be kept in the format set by the sender.

> Not necessarily. The storeage of data should be stored in whatever
> format is convenient, but presentation should use original formatting.

Yes, necessarily. How else can you verify the authenticity of a 
message?Reformatting for presentation on the other hand is not a 
problem.

>>> IETF doesn't generally define storage formats, rather formats used 
>>> for
>>> network transfer.

>> It seems rather silly to build a store-and-forward messaging system 
>> and
>> then only standardize the "forward" part and not the "store" part.

> The store in store-and-forward is about queueing. Not mailbox storing.

So what exactly is the difference between a queued and a stored 
message? I think that many people assume a message is composed, then 
goes through a series of queue/transmit cycles and is finally 
delivered. However, I see no fundamental difference between queuing a 
message for further transmission and storing it in a mailbox until the 
user is ready to look at it.

Another thing we need to look at is the difference between the 
person/entity who creates the message, and the person/entity 
responsible for sending it to the recipient. Obviously these are often 
the same, but for mailinglists and forwarded messages this isn't the 
case. In such cases, we need to be able to differentiate between 
replying to the original author, or to the sender. We probably want the 
sender to indicate which should be the default action, although this 
should be easy to override by the person who is doing the replying.