hello,

you might remember my bringing up this issue some time ago. here it is 
again.

i think that the qx.io namespace must be redesigned, and the qx.io2 
namespace should be unified with qx.io (or should be renamed to 
something else because it sounds too much like an ugly hack.) before 
freezing the api.

here are some issues, from my point of view:

1) qooxdoo must be able to support additional backend-communication 
protocols. if it ever does, the most elegant way to do this is to have 
classes like qx.io.rpc.Json, qx.io.rpc.Xml (xmlrpc) qx.io.rpc.Soap, etc.

You may have noticed my reluctance to use jsonrpc. it's cute and compact 
and all, but it lacks some important features like security, 
routability, authentication, etc. so i don't think it is up to par with 
today's enterprise requirements. So pushing JsonRpc as the one-size 
solution for backend queries renders qooxdoo less flexible than it is 
ought to.

Without taking the anti-rest propaganda seriously, you can read the 
following article with better explains reasons for my support of soap as 
a backend-communications protocol:
http://wisdomofganesh.blogspot.com/2007/12/paying-restafarians-back-in-their-own.html

i'd also like to note that the soap protocol has great server-side 
support -- you don't need to roll your own.

so, in short, i'm pretty sure that leaving room for other communication 
protocols will certainly add to the value of qooxdoo.

2) all those hypothetic classes should inherit from a common qx.io.IRpc 
a common rpc protocol interface. this will let us switch between 
different protocols with a minimum amount of effort.

3) I think that the qx.ui.table.Table is a great idea, well executed. 
Two points here:

   a) The "qx.ui.table.model" and "qx.ui.table.columnmodel" name spaces 
don't
      communicate their purposes precisely enough. i'd prefer to rename 
them to
      qx.ui.table.datasource and qx.ui.table.dataschema respectively. maybe
      better names could be suggested.

   b) The functionality of remote models could be applied to anything that
      scrolls. Comboboxes, Listboxes, and the like could benefit from such
      an abstraction. This also would make implementing an interface that
      "brings suggestions while the user is typing" a breeze.
i guess this sums my thoughts up. i could file those separately as three 
(four?) bug reports if you want me to.

awaiting your feedback

best regards,
burak



------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
qooxdoo-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/qooxdoo-devel

Reply via email to