Re: IMAP REPLACE extension

Chris -

Thanks for your feedback. The atomic requirement is intended to provide 
a guarantee to clients that use of REPLACE should simplify possible 
message states as compared to the existing 3 command sequence. I was 
really hoping to avoid a case where failure of REPLACE would require a 
client to retry by using 2 of the 3 underlying commands.

I certainly understand how it may be burdensome to some server 
implementations, though, so I'm hoping to hear from a few more server 
implementers before making a decision on whether to relax the atomicity 
requirement.

Thanks again...

- Stuart




On 5/11/15 10:03 PM, Chris Newman wrote:
> I believe the atomic requirement is too stringent and will reduce deployment of
> this proposal. A looser requirement can make implementation quite a bit simpler
> for some server designs. I'd state the requirement as follows:
>
> The REPLACE command is a combination of APPEND, STORE +flags \DELETED and UID
> EXPUNGE for a message. A response code for the STORE \DELETED flag MUST NOT be
> generated. The replaced and replacing messages SHOULD NOT be present in the
> mailbox at the same time.
>
> The difference with this phrasing relative to the stricter "atomic" phrasing is
> this does not require the server to eliminate the window where both draft
> messages are present in the mailbox. For certain server designs, eliminating
> that window can be either difficult to implement or significantly more
> expensive.
>
> The unnecessary "atomic" requirement deters deployment of multiappend extension
> as well. Slightly looser language is better for deployment. The MOVE command is
> a good example of language with appropriately loose requirements.
>
> 		- Chris

[snip]


_______________________________________________
imapext mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/imapext