On Tue, Oct 15, 2002 at 12:33:54PM -0700, Tom Lord wrote:
>...
> There _are_ open design issues -- but they are quite tractable if we
> can get moving on them.
Who is this "we"?
Tom, you have been doing a lot of talk about changing this or that, or
moving in some direction, or fixing bugs and releasing early, or whatever.
Unless *you* step up to actually start some coding on the ideas that you're
expressing, then I don't think you're going to get *any* traction here.
Speaking for myself and Karl, Ben, and Mike, we are focused on the state 1.0
plan which is going to put us a good ways past "45 days". I suspect that
many other people in this forum are also working against that plan. I
suspect that will continue to be the case until you plan to step up to the
plate and contribute actual code to demonstrate your ideas.
And back to "we": I personally plan to avoid a lot of the post-1.0 design
and planning. I also encourage Karl, Ben, and Mike to avoid post-1.0
planning in detail, but participation at a high-level is certainly fine.
(getting bogged down in details hampers our 1.0 efforts) There are a lot of
other people on this list, and your "we" can certainly encompass them. I
just want to set your expectation that when you say "we", that you'll only
peripherally be referring to the CollabNet folks. We've got a focus, and I
hope that we can retain that. Getting a *solid* 1.0 in people's hands is the
#1 priority.
[ regarding post-1.0, I believe we might look into putting together a
"Subversion Developer's Workshop" in December or January; we'll flesh out
the wrap-up of 1.0 and then apply forward-thinking to post-1.0 ]
>...
> This is, BTW, another reason to freeze the client earlier than you
> might have wanted to and factor out UI work to separate projects.
We are working under the ideal of a *feature* freeze for the client.
Improvements and polishing is certainly occurring, along with bug fixing.
Being an open source project, it is always a bit difficult to mandate a
feature freeze, but I think the community has been doing very well here.
"Big" features get discussed rather than "just going in." People seem to be
very cognizant of trying to avoid perturbation and getting that 1.0 out the
door.
> It's a big confusing mess trying to figure out what happens when you
> mix CVS-like file-at-a-time commits with changesets (you noted some of
> the issues in your example).
It certainly is not a mess -- we aren't trying to do changesets in 1.0.
>...
> but I tend to think future UIs are going to look a lot
> different from both the current arch UI, and the current svn UI.
If the future UI looks different, then hey... more power to it. But we know
what the 1.0 UI looks like.
> Both of the latter two will wind up being low-level, internal interfaces.
I think the command-line client will continue to be the single-most
important interface into Subversion. Over time, we'll see GUIs that bind
straight to the libraries (e.g. KDevelop, RapidSVN, and gsvn are all doing
this now!), but the cmdline will always be the most common UI.
But hey: if you want to see it as "low-level" and you want to code
"high-level" interfaces, then please be my guest. I won't and can't stop
you. But don't try to tell me or the other CollabNet folks that we're
somehow on the "wrong" path. I don't believe it, and neither do they.
Obviously, I don't communicate as closely with some of the other members of
the community here, but I suspect that belief is common. I would even say
that you've got to come up with some pretty incontrovertible proof to change
that view...
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Oct 15 22:37:19 2002