MAXHEADERSIZE value

Christoph Biedl cbiedl at gmx.de
Sun May 23 12:25:13 UTC 2004


Russ Allbery wrote...

> It's not supposed to be.  It's supposed to be the maximum physical length
> of any single *folded* header line (the number of bytes between CRLF
> pairs).

Ups. I have to admit that I didn't check the sources. But in the past
there have been a lot of articles with relative short physical lines (say
less than 80 characters) but several folded lines that caused the error
message and increasing MAXHEADERSIZE solved the problem.
One old example is
http://groups.google.com/groups?selm=p8qkj9.3m.ln%40tux.kursapbr.dialup.ruhr-uni-bochum.de&output=gplain
which was discussed (in German) in
http://groups.google.com/groups?selm=9krp6j%246b8nl%241%40fu-berlin.de&output=gplain

You see that this is an old topic for me but I cannot reproduce it in
2.4.x now - in 2.3.2 (Debian stable) the problem exists if the header is
"References:" or "Subject:", however a test article will pass with an very
long "X-Something:" header.

Seems like I'll have to dig the sources why this happens and whether this
is really still a problem in 2.4.x. If there was a major change to avoid
such problems, silently forget my initial message in that thread, thank
you.

> Where is it being used for the logical header length?  The logical header
> length should be completely unlimited.

Hm, this could become an interesting method to transmit binaries in
non-binary groups then.

	Christoph



More information about the inn-workers mailing list