Some interesting stuff from PLD
Russ Allbery
rra at stanford.edu
Mon May 5 18:12:53 UTC 2008
Julien ÉLIE <julien at trigofacile.com> writes:
> Manually browsing their CVS, I saw some patches which should perhaps
> be included in TRUNK:
>
> inn.conf
> --------
>
> -verifycancels: false
> +verifycancels: true
This option is pointless and loses legitimate spam cancels.
> -allownewnews: true
> +allownewnews: false
newnews is fast now and should stay enabled.
> -clienttimeout: 600
> +clienttimeout: 1800
Eh. I've never had ctlinnd take anywhere near the timeout.
> -innwatchspoolspace: 8000
> +innwatchspoolspace: 25000
This depends a lot on your system, but a larger value probably makes more
sense these days.
> -nicekids: 4
> +nicekids: 0
Eh.
> -docnfsstat: false
> +docnfsstat: true
Well, you don't want to run it if you're not using cnfsstat, since it's
pointless and just creates noise. But we could have it exit if you have
no CNFS buffers configured (it may even already do that).
>> +# note: must not be changed to false if using tradspool
>> storeonxref: true
>
> Why "mustn't" it be set to false with tradspool?
I'm not sure. I can't think of anything obvious off the top of my head.
> readers.conf
> ------------
>
> auth "localhost" {
> - hosts: "localhost, 127.0.0.1, stdin"
> + hosts: "localhost, 127.0.0.1, ipv6-localhost, ::1, stdin"
> default: "<localhost>"
> }
>
> I think it should be committed for IPv6.
ipv6-localhost sounds like a pecularity of a particular distribution, but
::1 should accomplish the same thing portably.
> Makefiles
> ---------
>
> http://cvs.pld-linux.org/cgi-bin/cvsweb.cgi/SOURCES/inn-asneeded.patch?rev=1.1;content-type=text%2Fplain
>
> It disables -as-needed for binaries using libstorage (there is circular libstorage<>libinnhist dependency).
>
> Something like that:
>
> -archive: archive.o $(BOTH) ; $(LINK) archive.o $(STORELIBS)
> +archive: archive.o $(BOTH) ; $(LINK) archive.o -Wl,-no-as-needed $(STORELIBS)
>
> Is it a good idea to commit that?
It's working around someone passing in an unsupported linker flag, so I
have mixed feelings about it; they're basically solving a problem that
they created. -as-needed is often a bad idea.
> configure.ac
> ------------
>
> -AC_CHECK_HEADERS(crypt.h inttypes.h limits.h ndbm.h pam/pam_appl.h stdbool.h \
> +AC_CHECK_HEADERS(crypt.h inttypes.h limits.h pam/pam_appl.h stdbool.h \
> stddef.h stdint.h string.h sys/bitypes.h sys/filio.h \
> sys/loadavg.h sys/param.h sys/select.h sys/sysinfo.h \
> sys/time.h unistd.h)
>
> Is it a good idea to commit that?
Well, it is if you don't ever want to support ndbm. I haven't used a
system with ndbm for so long that I don't have any real opinion.
--
Russ Allbery (rra at stanford.edu) <http://www.eyrie.org/~eagle/>
Please send questions to the list rather than mailing me directly.
<http://www.eyrie.org/~eagle/faqs/questions.html> explains why.
More information about the inn-workers
mailing list