[Seaside] Seaside with PHP (sort of)
marianomontone at gmail.com
Sun Apr 29 21:21:11 UTC 2007
2007/4/27, William Harford <seaside at harford.org>:
> On Apr 27, 2007, at 3:13 AM, Edward Stow wrote:
> Before investing too much time in the project you should have a look
> at some of the other php frameworks.
> The initial idea behind Phaux was simply to see if I could implement
> anything like Seaside in PHP. The proof of concept worked so well and was so
> exciting I decided to continue. Because Phaux is in PHP and not Smalltalk it
> lacks a lot of very useful Smalltalk features.
> I did look at other frameworks. None of them come anywhere near Seaside or
> Phaux in philosophical design decisions. All of them are traditional
> page biased frameworks.
We've implemented a component-based PHP framework. Have a look at it at:
The framework was developed at my company. It lacks community support and
I have a passing familiarity
> with CakePhp and am developing an application in Prado
> http://pradosoft.com. I suggest that you look at the QuickStart
> tutorial http://pradosoft.com/demos/quickstart/ .
> I have looked at Prado and considering the current state of frameworks it
> looks nice but ...
> Both CakePhp and Prado use code generation as integral parts of their
> design. I can't even being to tell you how disturbing I find this.
> Pardo uses templates. CSS is mature enough that templates should
> be relegated to a museum.
We use templates too, but they are declarative and optional. We don't
generate html programmatically. Template-less html is generated from widgets
which are some kind of "primitive" components. Html and XML for Ajax is
generated from the view tree.
No PHP framework other then Phaux is stateful (that I have been able to
PHPWebBuilder is stateful :). PWB is more stateful than seaside at the view
level. We mantain the tree of the view in memory. As a consecuence of both
being a tree and being in memory we are able to detect updates to the view
and send them through AJAX or Comet transparently.
Their mechanisms for persistence are, in my opinion, backwards. They
> generate (*shudder*) or base the object model off the relational model.
> REServe (both for Smalltalk and for PHP) do the inverse. Biasing the
> relational model off the object model.
We generate SQL from objects definitions too.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the seaside