inn-2.2 / storageapi + wireformat == corruption

David W. Hankins dhankins at mercenary.net
Sun Sep 26 22:01:38 UTC 1999


On Sun, Sep 26, 1999 at 12:36:08PM -0400, greg andruk wrote:
> The original QIO behavior was (mostly) correct.  To be pedantic it
> should actually have been canonicalizing (is that a word?) singleton
> CRs to CRLFs as well, but I do understand why that wasn't being done.

Ah, I agree but don't.

Molestation of the article body on articles that Have Badness(TM) is
perfectly reasonable and to some extent desired to protect news readers
and other news peers from those things a programmer does not expect they
would need to deal with.  In the CRLF case it can bring the article into
adhering with the standard with minimal risk to the article's core meaning
(which realistically was probably nothing thought-provoking anyway).

But I believe that if that is done it should be at INND time as the
article is received and stored on disk, not innxmit/other-backends time.
So I disagree.

Wireformat should mean on-the-wire format, not 'do-some-processing-first
format'.

With that assertion, I believe it's perfectly rational to expect that
two servers holding the same article in wireformat on disk have precisely
the same md5 sum over the article body and possibly some selected headers.

This is all rather moot of course since I believe the powers that be have
already decided that in the case of improper nul, \r and \n usage in an
article body, the article be forwarded as received.  This being thought
to be a 'better' behaviour in a bad situation than either dropping the
article or molesting it as you and I would have.

-- 
David W. Hankins	"If you don't do it right the first time,
Network Mercenary			you'll just have to do it again."
Currently For-Hire				-- Jack T. Hankins


More information about the inn-workers mailing list