The promised post.
First of all, I'd like to freely admit that I haven't been putting much
effort into Merb over the past few months -- certainly less effort than I
expected. The reasons are a bit complex and multilayered.

First of all, working on Rails is a tremendously complex endeavor. Virtually
every area Carl and I go into starts a monthlong (or longer) effort to
understand the codebase, do some refactoring work, and try to understand how
it fits into the long-term goals (which include a Merb transitional plan).
Most importantly, since January, we have been working basically non-stop on
refactoring/rewriting ActionController::Base. That work is not exactly
complete yet, but we did a lot more than we expected, at the cost of more
time than we expected.

If you look at Merb, it's mostly an analogue to Railties and ActionPack.
ActiveRecord, ActiveSupport and ActiveResource have no analogue in Merb. As
a result, a transition from Merb to Rails mostly involves a clear path from
merb-core to ActionPack. Specifically, Merb controllers need to both live
side-by-side with Rails controllers (this means the router needs to be able
to dispatch to them, for instance), and it needs to be possible to
reimplement Merb's controllers in terms of a stripped down version of Rails'
controllers. In current master, that means writing a Merb::Controller that
inherits from ActionController::Http, an extremely simple, fast version of
Rails controllers with opt-in components.

All the work we have done so far (and some we have yet to do) will make that
transition possible. We could have started releasing some transitional
releases over the past six months, but that would have been mostly
guesswork, and we likely would have needed several different releases,
requiring app
changes, as we continued our work in Rails. As Ezra said, Merb is
mostly stable at this point, and we'd like to keep it that way until
we can offer something clear and compelling. We have some more work to
do, but making it possible to transition smoothly remains a high
priority for the Rails 3 release for Carl and I.

With regard to the rest of the plans for Merb 1.1, some of them will likely
be making it into a release pretty soon, while others will need to wait a
bit longer. For instance, when we release a new router for Merb, we want it
to be the one that will be available in Rails. Since that work is still
ongoing, we're holding off on providing something unstable that will require
multiple changes to your apps. On the other hand, improved bundling can come
pretty soon; Carl and I will be working on app-agnostic bundling during our
work on the initializer and should have something that works with Merb. Of
course, bundling is a complex topic and we want to get it right.

I know it sucks a little to be in this holding pattern, but we absolutely
have not forgotten about Merb (for example, Solo/Flex is still a thriving
Merb app--feeling good about asking them to upgrade to a transitional
release is a big part of the criteria for doing such a release); we just
want to make sure that when it comes time for us to ask people to start
moving, we can do so with confidence.

Hope that answers some of your questions--the delay has been more about
shielding you guys from the vagaries of Rails edge than anything else.

-- Yehuda

On Sun, Jun 28, 2009 at 8:39 PM, Julian Leviston <[email protected]>wrote:

> Nice! :)
> It's a
> bit of a tip of the hat to merb, really... that it's taking a while to bring 
> Rails up to the integration point...
>
> Wickkkkked :)
>
> Julian.
>
>
> On 29/06/2009, at 12:49 PM, Yehuda Katz wrote:
>
> Again, there'll be more tonight, but we are ABSOLUTELY still planning on an
> easy transition path. Since we announced the merge, we've been working hard
> on cleaning up the internals of Rails so that such an upgrade path would be
> possible. As I said at the time, a transition from one API to another is
> relatively trivial, but a transition from a known API to the wilderness is
> not. Several times, we've looked at starting to write some transitional APIs
> for Merb, but realized we weren't far enough along in the Rails work to
> justify doing so without really dicking around the Merb users. I think we're
> finally far enough along to justify taking another stab at some transitional
> work, though. More details later tonight :)
> -- Yehuda
>
> On Sun, Jun 28, 2009 at 7:21 PM, Patrick Aljord <[email protected]> wrote:
>
>> The thing I'm more concerned is the 1.1 release we were promised and even
>> more important, the "easy" upgrade path to rails 3 we were promised by
>> releasing several versions of merb that would deprecate poco a poco the old
>> api to make it more simple to upgrade to rails3. At least that's what I
>> remember from the blog posts announcing the rails+merb merge but I might be
>> wrong.
>>
>>
>>
>>
>
>
> --
> Yehuda Katz
> Developer | Engine Yard
> (ph) 718.877.1325
>
>
>
>
>
> >
>


-- 
Yehuda Katz
Developer | Engine Yard
(ph) 718.877.1325

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"merb" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [email protected]
For more options, visit this group at http://groups.google.com/group/merb?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to