>>>> "Simon" == Simon Josefsson
<jas(a)extundo.com> writes:
Simon> Shouldn't people test a release before it is released?
Simon> Making to-be stable releases available on a branch (and
Simon> documenting that) seems like common practice. Only
Simon> mentioning the tag on xemacs-beta limits the audience to
Simon> stable release testers quite alot.
Not in my experience. The only problem I have is that people expect
to find unstable code on the stable branch. So they wonder why when I
say "cvs update -r r21-4-7rc1" and they do "cvs update" with a sticky
tag of release-21-4, they don't get the code.
For my purposes, it's plenty. When I say "please test r21-4-7rc1", I
typically get build reports for a half dozen platforms within hours.
If I could say "please test X, Y, and Z for a couple days", then a
more formal, "true beta", discipline would make a _lot_ of sense. But
I don't have a handle on that yet; "it built" is about as much
information as I can use.
Simon> If you get anything from CVS that isn't marked as a
Simon> release, you shouldn't expect it to work.
That turns out not to be the case. Those people who have questioned
the policy in the past typically (not all, but many) ask "why isn't
the `latest and greatest' available on the release branch?" Sounds to
me like they very much _do_ expect it to work.
And I think that's reasonable. If I make periodic requests for
testing, people are doing that as a favor to me---they should be able
to expect that the RCs are release quality, subject to the dumb
mistakes they're testing for. The rest of the time, building a
candidate-21-4 XEmacs is a crap shoot, and they know it. Or do now.
;-)
But if we advertise that branch on the web site, people who do _not_
know me or my work procedures will reasonably assume that
"candidate-21-4 is marked for release", and I _will_ get complaints.
"XEmacs didn't stop with eating my hard drive, it ate my lunch too!"
That's annoying to me, and really unfair to them.
I'll keep your comments in mind, but I tend to think that the "general
practice" is for the convenience of the core developers, not for the
users and beta testers. I'm not ready to change my mind quite yet.
--
Institute of Policy and Planning Sciences
http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Don't ask how you can "do" free software business;
ask what your business can "do for" free software.