[tpop3d-discuss] tpop3d, version 1.5.1

Chris Lightfoot chris at ex-parrot.com
Thu, 21 Aug 2003 11:49:01 +0100


On Thu, Aug 21, 2003 at 01:08:04AM +0200, Jakob Hirsch wrote:
> Chris Lightfoot wrote:
>=20
> > I have now bowed to the inevitable and released version
> > 1.5.1 of tpop3d.
> >
> >     http://ex-parrot.com/~chris/tpop3d/tpop3d-1.5.1.tar.gz
>=20
> looks very nice, as always. Especially the TLS support is really fine.
>=20
> some things:
>=20
> - tpop3d -h shows me compiled in mass-hosting though I
> set --disable-mass-hosting. hm, ok, just checked again, it should
> be --disable-masshosting, unlike --help says.

Thanks-- now fixed.

> - the man page seems a little outdated, e.g. it misses the -P option.

Also fixed. Not sure how that one slipped by.

(Thinking of which, there should probably be a timeout in
reading passphrases. I suspect I'm the only person who
actually uses key passphrases anyway, given the number of
programs which don't support a -P type option....)

> - (Unrelated to this particular release)
> auth_mysql seems not to be used at all if the username has no domain pa=
rt.
> Would you accept a patch that changes this? This may sound unuseful, bu=
t we
> have still many people that use there old unix username from
> sendmail_with_virtusertable-times. Right now I handle this with auth-pe=
rl
> but I'd like to kill this beast, so I simply can use an additional cond=
ition
> in the WHERE clause.

Umm.... I remember this behaviour, but I think it's now
been removed. Certainly looking at auth_mysql.c there's no
longer an explicit check for a domain being present.
Obviously if you specify a query involving $(domain) it
won't be run if no domain is given.

> (Note that append-domain is no clean solution. And I just compiled with=
out
> auth-perl for testing... wow, that saves about 850k making the binary o=
nly a
> third in size)

Well, perl is a bit bigger than tpop3d ;)

> - tpop3d.conf and the private key should only be readable by root, righ=
t? So
> what about some nice warnings (to stderr and syslog).

Noted.

> - IGNORE_CCLIENT_METADATA should be undefined if MBOX_BSD is undefined.=
 (any
> harm if not?)

Hmm. At the moment the only effect is to generate a
warning when compiling maildir.c. Your suggestion is
noted.

--=20
Two economists are walking along the street. The first says, ``Look, ther=
e's
a =A350 note on the pavement!'' The second replies, ``Don't be silly. If =
there
were, somebody would already have picked it up.''      (ancient economics=
 joke)