Yes, Christian, I am also looking forward to what you come up with.
Easier Said Than Done, IMHO.
"
Boy, when you get to my age you see good ideas die and reappear again.
"NoSQL" variable-length, multi-valued engines started with Dick Pick
(IBM 360 - 1965, then minicomputers(1973) and then hit the PC world in
the 1982 under the name "Revelation".  yes, it was a "revelation" to
those of us trying to build commercial applications on the lowly PCs
of those days.

Hundreds of vertical market apps and thousands of mission-critical
apps used these engines.  Their strength was is in their rapid
development ans prototyping capabilities.  No need for data
migrations.  (That's why I liked Hobo taking care of that for me.)

As Matt mentioned, reporting was the big problem. You had to build
your own (did that) or transfer to a "standard" database engine (also
did that) to satisfy enterprise-level clients.

Having said that, MongoDB has a natural synergism with Hobo. (But our
government clients are just getting comfortable moving from Oracle to
MySQL.)

I think this is a healthy discussion to have.  Dialectic is good.  But
who is going to do the experiments?  Resources are limited...

-Owen







On Feb 13, 6:50 pm, Matt Jones <[email protected]> wrote:
> On Feb 13, 2010, at 4:05 PM, Kristian wrote:
>
>
>
> > Some ambitious ideas for a new Rails 3 - Sproutcore framework.
>
> > Pink Palm framework
> > -------------------
>
> > Client - Sproutcore
> > Server - Rails 3
>
> > Client - Sproutcore
> > -------------------
>
> > Use Statecharts (Sproutcore) to handle state transitions in the
> > client, see Evin Grano.
> > Have a Ruby DSL to define/generate the Statecharts code. Each state
> > has a call to a javascript function with relevant parameters which
> > handles
> > the state transition! Generate Statechart graph from DSL using
> > Graphviz. Somehow the Graph generator should handle splitting a chart
> > up in multiple pieces. This could be assisted from the code side, by
> > naming state groups and then have the generator create a new graph for
> > each such state group.
>
> The first hit I got when looking for statecharts linked them to UML,  
> which triggered my "DO NOT WANT" reflex. :) Further digging turned up  
> the Sproutcore implementation, which bears a striking resemblance to  
> the way GUI apps have been in the past. I'd be interested in learning  
> more, but the sproutcore-statechart repo referenced in several posts  
> appears to be gone...
>
> > Define the view using Sproutcore Views. Community creates custom
> > Sproutcore views that can be used with themes.
> > Integrates various nice javascript widgets as Views, fx from jQuery UI
> > etc. See SCUI (SproutCore UI) project.
>
> Again, SCUI looks interesting - but I've got apps to develop *today*,  
> not in 6 months!
>
>
>
> > Server - Rails 3
> > ----------------
>
> > Pink Palm app. generator.
> > Rails templates for applying addons
> > Rails generators for adding Mongo models etc.
>
> > Rails app.
> > - Model: Mongo DB with MongoMapper/Mongoid
> > - Controllers: Rails 3 RESTful (Base)
> > - Routes: Rails 3
> > - Testing
> >  - RSpec
> >  - Cucumber
> >  - Capybara (replaces webrat)
> >  - Shoulda (optional)
> >  - Factorygirl/Mocha
>
> > Scaffold server app from YAML file:
> > Model, RESTful resource Routes, Controllers, Validations, Indexes and
> > Scopes (new Searchlogic gem?).
> > Have 2-way sync between YAML config and app code.
>
> Once again, an interesting idea - but why would I want to fool around  
> trying to say things in YAML that are better said in Ruby? Barring  
> massive intervention like grokking S-expressions into YAML arrays,  
> there's just too much that isn't expressible there: for instance, give  
> me a YAML representation of a named_scope with a lambda (which is all  
> of them, in Rails 3).
>
> A few points in summary:
>
> - the whole idea of configuring an app via a giant YAML file is  
> revolting. It smells way to Enterprise-Javay for my tastes.
>
> - Sproutcore is definitely something to watch, but there are some  
> serious barriers to using it in many apps:
>
> -- you end up repeating everything twice, once on the client side and  
> once on the server. The YAML configuration envisioned above appears to  
> be an attempt to address that, but there's still going to end up being  
> bits that aren't covered and have to be maintained twice as a result.
>
> -- as a corollary, the majority of business logic in the app is on the  
> client. If you ever wanted to watch C-level execs clench their  
> buttholes really fast, mention this. Rightly or not, there's still  
> some paranoia about competitors "stealing the app".
>
> -- the biggest issue facing JS client-side apps: IE6. It's still out  
> there, and will continue to irritate decent web developers for at  
> least the next few years. Between an incompatible and flaky DOM and  
> the 15x slower JS, it's a serious issue.
>
> -- the second biggest issue: the *back* button. No matter how many  
> times you tell users, "don't click back", they'll do it. HTML5  
> promises some relief for this with better state tracking, but see the  
> preceding point and add IE7's lack of support for HTML5 to the mess...
>
> ----
>
> In summary, I'd agree with the idea that richer, client-powered apps  
> are likely the future. Unfortunately, 90% of the businesses and  
> nonprofits out there aren't even running on *today*; big chunks of the  
> world are still being held together by Access DBs, giant cross-linked  
> Excel files and Word documents CCed to everyone in a department.  
> (Seriously - I just set up a Dropbox account for a small professional  
> association who was actually doing the CCed Word docs thing!)
>
> Anyways, I look forward to seeing what you come up with - despite the  
> rest of this post, *somebody's* got to get away from the PHP fumes  
> long enough to actually build the next wave of frameworks. :)
>
> --Matt Jones

-- 
You received this message because you are subscribed to the Google Groups "Hobo 
Users" 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/hobousers?hl=en.

Reply via email to