On Sun, 2006-02-19 at 22:59 +1300, Simon Kitching wrote: > On Sun, 2006-02-19 at 07:59 +0000, robert burrell donkin wrote: > > On Sat, 2006-02-18 at 16:12 +0000, Niall Pemberton wrote: > > > On 2/18/06, robert burrell donkin <[EMAIL PROTECTED]> wrote: > > > > --8<------------------------------------------------------------------- > > > > [ ] +1 Approve this process > > > > [ ] +0 > > > > [X] -0 > > > > [ ] -1 Do not use this process > > > > ----------------------------------------------------------------------- > > > > > > I think it would be a mistake to label it as "alpha". In my mind this > > > implies either an API that is unstable and subject to change and/or > > > software which isn't fully finished. As such many people may ignore an > > > "alpha" release and it could reduce the number of people who test it > > > out - since we want it tested as widely as possible my vote would be > > > that it is labelled "beta" once it "satisfies the quality standards > > > required for jakarta commons releases", rather than "alpha". > > > > i had it in mind to release the alpha without a major announcement > > outside the commons. we need to approach some important downstream users > > and re-packagers (jboss, axis, tomcat etc) with the new code. a beta for > > public consumption and testing would follow once the alpha's been > > checked by developers for those projects. > > > > but i can live with going straight to beta if that's what people think > > best... > > Sorry but I don't like the idea of calling the code "beta" either. I > would prefer to tell Tomcat/jboss/axis etc that there is a JCL 1.1 RC5 > available. That tells everybody that we think things are very close to > releasable state. It also tells people that they need to test & provide > feedback pretty quickly if they want to influence the release. Calling > this a "beta" release implies neither of those things.
it's not about influence: it's about ensuring that the code base we have really is compatible and really solves the problems we think it does. i *really* don't want to spend the next year of my life fielding abuse about how bad JCL 1.1 is. IMO the only way to ensure that this does not happen is to get the code well tested in containers before it's full official release. using alpha and beta's for testing is well established amongst apache communities (and others) with large existing installation bases. a release candidate is not an official distribution of any kind. it's very difficult to get unofficial distributions tested by anyone who is not a developer of the product. it is very difficult to generate any kind of publicity around an unofficial release. IMHO it's going to take a lot more effort to talk organisations into testing an unofficial distribution that it would take to approach them about testing compatibility with an official public alpha or beta release. - robert --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
