Sis attachment deduplication

Reindl Harald h.reindl at thelounge.net
Fri Jun 20 11:23:06 UTC 2014



Am 19.06.2014 16:52, schrieb Laeeth Isharc:
> 1. Is this now reasonably stable for large mailboxes (c 2mm messages)?

dunno

> 2. Will this leave the filename in the message body unchanged?  So for example if I have the same attachment called proposalfromvendor.pdf and proposaltoclient.pdf in two different emails, will the original names be kept ?  Or will it replace the filename with some kind of numeric hash ?

the message body must not be changed
never, not in any single case for no reason

why?
because it breaks all sort of sigend mails

dunno how this is implemented, in case of dbmail each message
is splitted in it's mimeparts, they are all stored with a sha1sum
and referenced to the messages, fetch a message re-constrcuts the
whole mail from it's mimeparts and everytime a new mimepart arrives
it's verified with the existing checksums to decide if a new record
is needed or only a reference - after all references are gone the
record for a mimepart is deleted

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 246 bytes
Desc: OpenPGP digital signature
URL: <http://dovecot.org/pipermail/dovecot/attachments/20140620/3ec54a92/attachment.sig>


More information about the dovecot mailing list