On 8/1/11 01:10 , Paul Merlin wrote:
I don't have anything usefull to say about IDE plugins as I don't use
them much.
I could live well with any good java code editor alone (classpaths,
autocomplete).
Actually, and just for info, I still use Netbeans to write Qi4j code
reusing the
poms generated by gradle to get the IDE classpaths rights. It works for me.
When it comes to IDE plugins, my main concern is to get Envisage type
functionality in there, similar to the Gradle plugin which allows you to
visualize tasks. I would want to be able to code assemblers and
composites, and then hit "Refresh" and see in the visualizer that it
looks like what I expected.
At my shop, software editor, we use a Qi4j app embedded in a JEE app.
Each one
of them have its database. All this is using XADataSources managed by
the JEE
container. DataSource are wrapped before being used by the SQL
EntityStore so
that's the JEE container (or JEE app code) responsibility to issue the
commits.
The only limitation we have is to play one UoW only per request and that
suited
our use cases. The final product has backup/restore and HA facilities
eased by
the fact that all storage is done in SQL databases. Today it is used
successfuly
in production at several places with tens of thousands users.
Right, so this is what I was getting it. I.e. not necessarily the "SQL"
part, but more the production part. Whenever I deploy Streamflow the
admins ask for the db schema, and when I say "there is none" a long
debate ensues. Avoiding that would be nice.
Also, I'm looking at multi-tenant hosting options (i.e. one WAR, many
Qi4j instances, each with its own settings), and with that it helps to
store data in a centralized database with one logical DB per customer.
Backup/replication happens on the whole thing.
About Validation I guess that would mean something like being able to
use JSR303
based annotations to apply Qi4j Contraints. Right?
Basically.
About Injection, would that mean supporting at least the @Inject/@Named
annotations and the Provider type?
Basically.
I'm personnaly convinced (mostly by you Rickard :) ) that JSR330 is
wrong and I
don't know if supporting it would ease the learning curve.
So that's my 0 cents on this one ...
Yes, it *is* wrong, but AFAICT it would be reasonably easy to support.
Not sure if it helps or hinders though...
I tried to bring some 'new blood' to Qi4j but despite the fact that all
of them
told me "there's a lot to read", being happy with that, only a few got a
good
grasp on the technology because they missed both big picture tutorials and
getting started tutorials taking them by the hand from getting the sdk,
setting up their classpaths rights to a running example. The "application
templates" could be a great source for good tutorials.
Yeah, I think so too. If you can avoid having to invent a good layered
architecture, and instead "get to work" quickly, given that you've
chosen a decent template, that would probably help.
* Getting a full stack. The main missing point to get a "full stack",
as it is, is the REST part. I have this, and can port it to Qi4j, so
that should be doable. Once that is in place we have a good end-to-end
stack for building apps, I think.
I'm really interested on this one, I'm building REST webservices using
Qi4j and
I'm sure I already reinveted some of your wheels there.
With REST, AFAIK the REST library I did for Streamflow is the only
framework currently available that properly supports HATEOAS and the
other ideas that Roy was talking about. This is sooooo missing in e.g.
Jersey et al.
So those are my thoughts on this. Is it doable? Or should I give up
on this? Is it possible to get any meaningful "marketshare" with Qi4j,
if v2.0 has all of the above? What is important? What is not
important? What do you guys think?
No please, don't give up, Qi4j is important!
I'd be happy to help by working on extensions and libraries.
I'm working on an opensource project using Qi4j, maybe it could give some
visibility to it once/if it get some. If projects using Qi4j get
visibility,
Qi4j itself get some of it. What about a PoweredBy page on the site and a
"PoweredBy Qi4j" official logo? BTW, has the StreamFlow project been
finally
opensourced?
That would be nice! As it is, when people ask me about how many are
using Qi4j my answer is necessarily "I have no idea". I myself don't
know if this is all a waste of time, or if there is real interest in
these things from others than myself. Any feedback you can give here
really really helps.
thanks, Rickard
_______________________________________________
qi4j-dev mailing list
[email protected]
http://lists.ops4j.org/mailman/listinfo/qi4j-dev
