Alex Baer wrote:
Rich Gray wrote:
Alex Baer wrote:
Re-creation of index files was unfortunately not done instantly in
earlier versions. For large mbox files it takes ages. The whole concept
slows down actions, such as copying 500 messages from one folder to
another one significantly. It takes so long, at times, that I guess, the
index is recreated for every single message. May I expect that Seamonkey
has become a bit smarter here in recent incarnations?
Indexing and copying operations will be slower with mdir. For mbox,
only a single file per folder must be indexed. With mdir, each message
is a file, so in your 500 message folder example, there would be 500 file
opens and closes. That's a heck of a lot more overhead. Same for the
copy. 1000 opens and closes. Mbox should be much faster for such things.
In my mind, the advantage for mdir is with incremental backups, like
Apple Time Machine and search mechanisms like Spotlight. Receiving
a new message means that only that new file need be backed up, instead
of the whole folder's worth of data. Presenting messages to Spotlight
is hard in the many messages/file mbox format.
Is there really an index for maildir needed?
I would think you'd need it more, if dealing with large numbers
of messages. Instead of accessing a single file (the index or
mbox), you'd be accessing one file per line in the header
display. Could get to be a pig for sorts and searches.
I assume, maildir would work
quite well without an index, because each message is a file, and every file
has an "index" in the file system.
The only information one can depend on to be portable would be file things
like name, size, creation timestamps, etc. Even if you can stuff all that
header information into file system extended attributes, I would guess it
will still be significantly slower than opening a single file and blasting
through a single chunk of data at CPU speed. Without an index, maildir
would seem to require a filesystem access per message.
In fact, I don't see index files in maildir filesystem folders.
I don't know how other apps organize their data.
Also, I would assume, that reading and processing multiple message files can
be done in parallel on modern systems, while mbox probably only allows
sequential processing.
They will still collide at the filesystem/disk.
Not sure, if mail clients supporting maildir really
read and process multiple messages in parallel, though, frankly.
Regardless, if my assumptions above are correct, my experience is, that
KMail is quite a bit faster working on folders with many thousands of
messages than Seamonkey Mail. But I have to add, that operations where this
speed difference becomes evident, are rarely used, such as copying large
bunches of mails between mail folders. So it's noticeably, but not something
that would prevent anyone from using Seamonkey Mail.
For "reasonably" sized mailboxes, I'm not sure it would be terribly
noticeable either way.
--
Rich (Pull thorn from address to e-mail me.)
SeaMonkey - Surfing the net has never been so suite!
_______________________________________________
support-seamonkey mailing list
[email protected]
https://lists.mozilla.org/listinfo/support-seamonkey
