On Mar 8, 2007, at 5:23 PM, robert burrell donkin wrote:
On 3/8/07, Alex Blewitt <[EMAIL PROTECTED]> wrote:
On 08/03/07, Geir Magnusson Jr. <[EMAIL PROTECTED]> wrote:
>
> On Mar 8, 2007, at 2:27 PM, Alex Blewitt wrote:
>
> > We ought to have a version monkier as well as M1, to ensure we
don't
> > get confused in the future. Apache Harmony 1.0M1 would make
sense, or
> > if we want to sync with the proposed Java versioning numbers then
> > Apache Harmony 5.0M1 might make sense.
>
> Please not 1.0M1. Please please please.
Why?
another reason: once you start publishing 1.0XXX releases, the world
outside tends (wrongly) to think that harmony is at 1.0
That describes it fairly well for me. I've experienced this in other
places, and tagging anything w/ a 1.0 sets an expectation.
I'm far happier to manage expectations than have them managed for me,
so to speak.
For me, I'd put the 1.0 on it when we have an "alpha", or "beta" or
"rc1" - when we're doing TCK, have a handle on what needs to be done,
and getting close to seeing 'all green' in the TCK tools.
For now, I can live w/ fractions (0.5), but there is a problem there
- I've seen projects that did that and got onto what I jokingly refer
to an asymptotic approach to 1.0, such as
0.91
0.915
0.92
0.92333444222
:)
I think castor had that problem. Always "almost there".
Unlike something like, say, Apache Velocity, where we could decide
"Aw hell, lets call it 1.0", we have a hard boundary at the moment
before we go 1.0, namely Java SE 5 compatibility. So I want to avoid
being forced into naming contortions. That's why I proposed names.
geir
