Re: a Z-mailer question

[mea@nic.funet.fi]

[ Sorry Roy, you asked this privately, but replying on this does cover
  several questions I saw in my inbox this morning ]

> Hi Matti:
> 
> I have installed your latest Z-mailer snapshoot on a Solaris 2.6 box and
> everything is fine except now $orgdomain stuff doesn't work like before
> (i.e. in release 2.99.49p9).  If I address mail to this system from the
> one I installed the latest release on, it will not work as
> 'rcb@press-gopher' but only as 'rcb@press-gopher.uchicago.edu'

	(this question keeps surfacing every so often, nearly half
	 of all questions I see are related to this one..)

	Pardon me, but when did you folks stop following the INSTALL
	instructions about `localnames' database ?

	While speaking of   $MAILVAR/db/localnames  the file tells:


	REMEMBER: ALL NAMES THAT THE HOST MAY EVER HAVE ARE BEST LISTED
		  IN HERE!  It reminds you of them, and makes sure a message
		  destined into the host really is accepted.

	I think it is very much same as with sendmail Cw/Dw rules, though
	that of course gets also `hostname' derived identities there auto-
	magically, while  ZMailer's  `localnames'  file does not have them.


	For the router that is not quite strictly mandatory, but...

> I take it, judging from the warnings about 'unused variables', that you
> are putting even more junk e-mail protection in there.  Oh, goody! :-) 

	... that file is the place from whence the   policy-builder.sh
	script pulls in its knowledge about system local identities.

	Those codes you refer at are from -- damn, I haven't added credits
	and the patches didn't have them to begin with. (hint, hint!)
	Now whose code it was ?  I think they came in from Russian.

	I am planning to write a generalized content sensor program
	interface -- it is already at the CVS tree very least.
	( See  smtpserver/contentpolicy.c  file. )

	Now it is an empty stub, though, as I haven't decided on how
	to drive the external program.  It could be something forked
	below smtpserver session much like router can be forked, but
	it could equally well be an AF_UNIX/AF_INET communicating
	external server program which is called with some sort of RPC
	call and that does sensorship analysis for all smtpservers.

	Anyway, it shall return a non-zero return status for each call
	when the message has something questionable in it.  It shall
	yield negative values when the message is absolutely rejected,
	and yield positive non-zero values when a message is sent into
	"freezer".  At the time of the call the spool-file is completed,
	but it has not yet been moved to the router directory.


	I am trying to slow down the avalance of new code -- haven't done
	much of anything for a month now, except at router scripts.
	Instead I have been doing my work-work, and also spent some time
	at trying to get the documenting back on track.

	Anybody out there who knows how to get  LaTeX2HTML  to behave like
	I want it to be ?  (Oh, the LaTeX works, but when I want to create
	multipart HTML material out of the same, it begins to loop...)
	It is synced in the CVS tree -- doc/manual/, and you need L2H
	package version 98.1p1 (the lattest)  (And xfig 3.2!)

	We made a lot of work by using LinuxDoc-SGML only to finally realize
	that for a book it makes lousy looking results, nor the HTML it could
	yield was that smart..  In about a day I was able to turn that SGML
	produced LaTeX into LaTeX2HTML sources, and with LaTeX it does behave
	properly, but to yield HTML I am stymied..

> ---
> Roy Bixler
> The University of Chicago Press
> rcb@press-gopher.uchicago.edu

/Matti Aarnio <mea@nic.funet.fi>

PS:	While many people believe that Finns are born with skis in their
	feets, it means cross-country skiing.  Two days ago I had my first
	touch on downhill -- alpine -- skiing, and now I am sore all over...