Paul J Stevens wrote:
Michael Monnerie wrote:
On Samstag, 16. Dezember 2006 14:52 Leonel Nunez wrote:
http://www.postgresql.org/docs/8.2/interactive/datatype-binary.html
Thank you, found it already - I was looking at strings.
Back to the topic:
Although I prefer when software "just works", but wouldn't we possibly
have other problems when accepting utf-8 headers? After all, they must
not exit, and I believe postfix won't accept them already (Wietse is
very strict on rules).
Being able to accept utf8 headers would be a side-effect, not a design
goal of a BYTEA conversion.
One issue is headers can contain modified utf7 headers which encode 8bit
data. How to deal with that from dbmail's perspective: read IMAP-SORT
and IMAP-SEARCH. These won't work against utf7 strings so some kind of
decoding is required. The problem is when and how.
So those broken headers come from broken software, would it be wise to
support that? If it doesn't create other problems I don't care, but I
can imagine somebody sending a header with a NULL byte (\000), breaking
SpamAssassin and such.
I suspect the original report by VladK is actually about two issues:
incorrect 8bit encoded headers, plus incorrect parsing of the message by
gmime. The latter could well be the result of the former. I don't know.
I am planning to run SA over already received e-mails, to find SPAM
afterwards - and that could crash, or spammers could hide their domain
names, e-mails etc. behind such a \000 byte...
I wouldn't worry about that. I'm pretty certain gmime doesn't accept
\000 bytes either.
Remember, we are discussing Postgres. Why not create a domain based on
bytea with a constraint that prevents \000 and any other undesirable
characters?
_______________________________________________
DBmail mailing list
[email protected]
https://mailman.fastxs.nl/mailman/listinfo/dbmail
