Perhaps just simplicity?
I think we want the schema to be 'honest' as Don says. And if it
turns out that an honest schema is intractable, that may be a hint
that we need to adjust our class implementation. It's never been
clear to me why canvas is implemented as a view, since it works real
hard to subvert a lot of view initialization, since it can't really be
a view. Perhaps someone thought they were being clever because canvas
wanted a lot of properties that happen to be just like view? Perhaps
there needs to be a refactoring where the bits that are shared by
canvas and view are factored out into a common class?
On 2008-05-23, at 15:53 EDT, Henry Minsky wrote:
I forget why canvas wasn't supposed to be a view...
On Fri, May 23, 2008 at 3:30 PM, Donald Anderson
<[EMAIL PROTECTED]> wrote:
In lfc.lzx there is:
<interface name="canvas" extends="Instance">
but it looks like <canvas> really extends <view> ?
Likewise there are many attributes listed on canvas, like:
<attribute name="bgcolor" type="color" value="white"/>
which is really an attribute of <view>. I guess this is historical?
I assume we want the schema produced to be honest about this,
or else what?
Also there are are these tags:
<interface name="canvas" extends="Instance">
....
<containsElements>
<element>view</element>
<element>script</element>
These are the items allowable within a canvas, right?
I don't think we have a way to deduce that presently.
Do we want to list these somewhere in the javadoc,
for each class, or maintain a separate lzx that gets merged in,
or ...?
Which would be easier to implement and maintain? I guess comments in
the
class source files would be easier long term to maintain.
On May 22, 2008, at 4:34 PM, P T Withington wrote:
Yes, the task is to compute the <interface> declarations that
represent
the API of the LFC.
On 2008-05-22, at 16:23 EDT, Donald Anderson wrote:
Tucker,
I'd like to get started on this one:
http://www.openlaszlo.org/jira/browse/LPP-3508
Can I assume that the contents of the schema/lfc.lzx file
represent all
interfaces
compiled via WEB-INF/lps/lfc/LaszloLibrary.lzs ? That is, all
lzs files,
and no lzx? If so,
I think the best approach is to have an option to js2doc to do
this.
- Don
--
Don Anderson
Java/C/C++, Berkeley DB, systems consultant
voice: 617-547-7881
email: [EMAIL PROTECTED]
www: http://www.ddanderson.com
--
Don Anderson
Java/C/C++, Berkeley DB, systems consultant
voice: 617-547-7881
email: [EMAIL PROTECTED]
www: http://www.ddanderson.com
--
Henry Minsky
Software Architect
[EMAIL PROTECTED]
