At a minimum if we have passed the bugzilla threshold, my next criteria is I
generally don't like seeing a release go out that has known make test
failures.
Currently, I'm failing on this:
t/dkim2...........................ok 32/123 Not found:
DKIM_ADSP_NXDOMAIN = DKIM_ADSP_NXDOMAIN at t/dkim2.t line 85.
# Failed test 120 in t/SATest.pm at line 716
Turning on verbose tests, I found this output:
Checking for anti-pattern DKIM_ADSP_ALL at t/dkim2.t line 85.
*** Mail::DKIM 0.36, Test 120 is expected to fail with versions older than
0.36_5
ok 121
I'd like to either A) see that warning in non-verbose tests, B) Makefile.PL
modified on line 185 to say 'Mail::DKIM' => 0.36_5, or C) see a warning
from the Makefile.PL advising they upgrade if not require it before I +1.
So consider me a +0 at best right now.
After upgrading to 0.37, I've confirmed that the test passes and all other
tests with make tests with net_tests enabled looks good.
For some reason, I seem to remember a previous recommendation similar to
this so not trying to beat a dead horse if I did and it was quashed.
re: new names, I'm fine with every release being separate. I'd even like to
see the dir it untars default to a suffix of -beta1, -beta2, etc.
Regards,
KAM
----- Original Message -----
From: "Warren Togami" <[email protected]>
To: "SpamAssassin Dev" <[email protected]>
Sent: Thursday, December 03, 2009 3:58 PM
Subject: PROPOSED: Apache SpamAssassin 3.3.0-beta1
According to build/README we can release this pre-release with "lazy
consensus" if nobody objects. Please test this tarball. If nobody finds
anything horrible by Friday morning we will bless it as beta1. I've
personally been running this in production since last night, but it took
me until this long to figure out how to sign the tarballs.
(I also think we should stop re-cutting pre-releases using the same name
but different tarball contents like the past, if we have to re-cut the
beta. The next tarball will be beta2 if we find horrible problems with
beta1.)
