Subject: [netatalk-admins] fixing papd - pre-patch 1
From: Andras Kadinger (bandit@freeside.elte.hu)
Date: Thu Sep 03 1998 - 19:30:51 EDT
So, I came up with this a couple of days ago, I just didn't want to send
it out for I hoped I can do overhauling of papd in a reasonable time.
This gross hack modifies the line marking routine in papd to handle both
the one and two character version line endings valid in PostScript, and
hopefully also removes 'canonization' of them so that incoming data goes
out unchanged.
This however doesn't fix the logic in papd, that forces it to read an
'entire line' (that is, up to a lineending) regardless of the length of
the line - so you should still be able to exhaust all your virtual
memory (and eat up enormous CPU time for that matter) by sending a large
binary job to papd.
One possible workaround could be here to let markline return a long but
unfinished line as a complete one, preventing papd eating up memory.
This however has some chance to disturb comment handling; but since papd
currently handles only query comments necessary for spooling to work,
the best would be here I think to make up some heuristics to recognize
the beginning of the actual print job, and blindly copy data to the
spool from that point on. Does this sound plausible?
Andras
This archive was generated by hypermail 2b28 : Sat Dec 18 1999 - 16:33:14 EST