[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: IMAP capability for maximum APPEND message size?
On Wed, 2005-11-23 at 16:08 -0800, Mark Crispin wrote:
> I am considering the addition of a trivial new capability to IMAP that
> would advertise the maximum size of a message in APPEND (and probably also
> CATENATE). This has the same purpose as the SIZE capability in SMTP.
> Currently, the only message size limit in IMAP is 4294967295 (2^32 - 1)
> due to IMAP's use of unsigned 32-bit integers. Supporting that size is
> troublesome due to +1 overflow problems. Many implementations have much
> smaller limits, such as both for administrative and technical reasons
> (many filesystems encounter problems after 2^31 - 1).
> Rather than have this be a secret, I propose a capability called
> MAXAPPENDSIZE=nnn, where nnn is the size.
Well, personally I don't see a need for it. If you're sending message
size in APPEND as a normal literal, the server can reject it
immediately. So it would be useful for literalplus only.
And is it really the message size limit that's a problem? If your
system/code is able to handle only 2GB files, how does 10MB message size
limit help if the whole mailbox size is already 1,99GB?
Also I think in many (most?) cases you'll reach out-of-quota condition
before server's maximum allowed message size, and in that case it
doesn't matter what the message size limit was.
Is +1 overflow problem really an issue? If you're using 32bit file
offsets, the file limit is 2GB. As long as you're using unsigned
integers to perform your internal calculations, there are no overflow
problems. If you want larger files, you're using 64bit offsets and again
there are no overflow problems.