> It's probably about time for another release, too.
> There haven't been any major changes, just a handful
> of little things, but it adds up.
This proves one thing: Maverick is well designed and very mature
framework.
One thing comes to mind. I just throw this on the air. Have you thought
about command chaining? And, if yes, what have you decided? What I mean
about chaining is that you could build independend reusable commands
with it. Or is it already possible? If yes, then how does it work?
client side redirection is not what I'm looking for.
Simple example:
- User signs guestbook form
- signGuestbook.m command is executed
- signGuestbook command's Sign controller is called
- Sign controller stores guestbook entry to db
- Sign controller returns SUCCESS
- SUCCESS maps to another command -
retrieveEntries (in this case not a view)
- retrieveEntries.m command is executed
- retrieveEntries command's Retrieve controller is called
- Retrieve controller get's guestbook entries from database
- Retrieve controller returns SUCCESS
- SUCCESS maps to a view
- View is displayed to User
There is many things that need to be noted about this:
How is the model for a view build in a command chain?
- The last command in chain builds it?
- Every command in chain can add itself or
something else in the model, but how?
- Following actions could somehow extend the model
(extensible Model interface or something like it)?
- Approach like the one in Webwork, i.e. ValueStack?
- Support for multiple models (?)
- When the model is build? And is the model that
is build in previous commands usefull in following ones?
- How to implement Model lifetime in a chain?
- Add your questions here!
How the parameters are passed to commands in a chain?
- Every parameter is copied to every command if it
has specified property in a form bean or controller?
- Parameters are only passed to the first command?
- Can command that are prior other commands change
the parameters?
- Add your questions here!
Other things:
- Should there be parameter with default value
that limits the number of commands that can be
in a chain to protect from infinite loops?
- Then throw an exception?
- Should we think about some Workflow engine integration?
- If yes, should we write documentation about it?
- Is this too much bloat?
- What happens when exceptions occurs in a chain?
- Should this framework care about exceptions?
I think that struts has something like this,
but I have not used it since v. 1.0.
- This thing is not releated to chaining itself,
but the general framework.
- Add your questions here!
- Is there some other approaches for this?
- Controller extending controller extending controller...
- super.perform()
- Not independend
- Not very reusable
- Problems with parameters (i.e. same name)
- Workflow engine? What do you recommend?
- Views could call multiple actions with different
parameters?
What do you think?
> I revised a bit of the documentation, too.
Great!
> Scott, Jim, a release of 2.1.2?
Go for it! Althought I'm not Scott or Jim, :-).
Kind Regards
Aapo <b/>ungle Laakkonen
-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
[INVALID FOOTER]
