[Laszlo-dev] Re: serverless operation toplevel attribute
Adam Wolff
adam at laszlosystems.com
Wed Jan 19 09:54:17 PST 2005
I can easily accomodate this. I still think that I should be able to
make media requests bypass the proxy on a request-by-request basis.
Consider the case where I have some gif resources and some jpg
resources. I need the server to transcode the gifs, but I would
probably want to bypass the server for the jpgs.
I also think that overloading the 'cache' argument to setSource(and
renaming it) is the best entry point for this.
Adam
Oliver Steele wrote:
> [Moving to laszlo-dev:]
>
> On Jan 18, 2005, at 6:39 PM, Henry Minsky wrote:
>
>> The hack for 'serverless' http data sources right now is that if you
>> construct a dataset with the "transport=direct" attribute, then
>> LzDataset constructs an anonymous XML datasource as its parent,
>> otherwise it constructs an LzHttpDatasource.
>
> Here's how to bring this into line with the Compilation Switch+Toplevel
> Attribute proposal:
>
> - Add <canvas proxied="false|true|inherit"> to the schema.
> - Add LzCanvas.proxied. Change CanvasCompiler to initialize this from
> <canvas proxied>, when it is present.
> - When the application is initialized, if canvas.proxied=='inherit'
> then set it to true or the value of the lzproxied query parameter.
> (I'm thinking that the query parameter should be lzproxied instead of
> proxied, as per "fake query parameter namespaces".)
> - Modify LzDataset.prototype.construct to test for (canvas.proxied ||
> args.transport=='direct').
>
> Can something similar be done for media requests?
>
> This implements the parts about libraries, compilation warnings, or
> debug warnings, but it gets enough of the feature implemented to begin
> using it. It also leaves <datasource direct> in the schema; we can
> remove or rename that once we have a final spec.
>
>> There is actually at present no LZX tag to construct an
>> LzXMLDatasource from script.
>>
>> This implementation needs to be changed, although, does anyone
>> actually use <datasource> in their code anymore, or do they just
>> declare <dataset> for everything?
>>
>>
>> LzDataset.prototype.construct = function ( parent , args ){
>> //Default datasource type is http
>> if (! parent.instanceOf(_root.LzDatasource) ) {
>> this.oncanvas = parent == _root.canvas || parent == null;
>> if (args.transport == "direct") {
>> parent = new _root.LzXMLDatasource( _root.canvas,
>> { name: "anonXMLDS" +
>> _root.LzDataset.anonDatasourceNum++});
>> } else {
>> parent = new _root.LzHTTPDatasource( _root.canvas,
>> { name: "anonHTTPDS" +
>> _root.LzDataset.anonDatasourceNum++});
>> }
>> parent.repset = true;
>> }
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Laszlo-dev mailing list
> Laszlo-dev at openlaszlo.org
> http://www.openlaszlo.org/mailman/listinfo/laszlo-dev
More information about the Laszlo-dev
mailing list