Re: mailbox format(s)

Iljitsch van Beijnum <[email protected]> Sat, 28 Feb 2004 15:15:35 +0100
Newsgroups gmane.mail.ng
Message-ID <[email protected]>
On 28-feb-04, at 3:06, Bruce Lilly wrote:

>>> Not with IMAP, where the message store typically lives on the
>>> server

>> Well, then a new client would have to download all the messages from 
>> the
>> server again, wouldn't it?

> No.

Ok, it seems we're not communicating. I assume the client would want to 
have a local copy of all messages, you seem to assume that the client 
wouldn't want to have local copies of any messages. Regardless, if you 
use IMAP then the fact that MUA #1 has downloaded a message doesn't 
save MUA #2 from having to download it too if the user wants to look at 
it.

> There is rarely
> any reason to examine every byte of every file using IMAP with
> the type of storage described (one file per message).

What you're saying has no bearing on my objection that using a file per 
message takes much, much more disk space than aggregating messages into 
a smaller number of files.

>> Yes, *today*. With a system that includes message sizes the body can 
>> be
>> skipped

> IIRC, that's been tried ("Content-Length" stored in the message
> header), and it has led to corrupted mailboxes.

This is all physical database design 101, there are people around who 
are actually capable of implementing something like this correctly.

>> It would be pretty stupid to store messages such that an operation is
>> done once a month can be performed easily while operations that happen
>> every minute suffer.

> What is the basis for your statement that "operations that happen
> every minute suffer"?

Having to store binary data in text files as opposed to text in binary 
files means that all access to all data, but especially binary data, is 
less efficient. I'm assuming people are accessing data on a mail server 
every minute.

>> Anything that displays the text between the tags is a valid HTML
>> implementation.

> You're joking, right?

No joke. Anyone who writes HTML and makes assumptions about what will 
be displayed other than the text between the tags in some form or 
another lives in a fantasy world. Why do you think so many websites 
specify the user agent to be used for visiting them?