Tim Schaub wrote: > Hey- > > Eric Lemoine wrote: >> On Sat, Feb 23, 2008 at 9:49 PM, Sean Gillies <[EMAIL PROTECTED]> wrote: >>> Eric Lemoine wrote: >>> > The OpenLayers Ajax code doesn't allow for other HTTP methods than GET >>> and POST! >>> > >>> Is anybody currently relying on this feature of Ajax.js? >> "Feature" ?? ;-) >> >> FYI Tim Schaub has started working on new XHR support in OpenLayers. > > Yeah, I recognize I sound a bit like a broken record, but I want to get > out from under this hacked ex-Prototype.js mess. > > I mentioned the POST/GET problem (limit to those verbs) to Chris a bit > ago as another reason to ditch the whole thing. I don't see value in > continuing to fix this code (though that one is an easy fix). > > So, I've put together a proposal for what I think is a better solution: > http://trac.openlayers.org/wiki/Proposal/Request > > (Note that all code on that wiki page and in the sandbox mentioned there > was written without testing. I'm confident that it will all work in the > end but will not be surprised if someone points out typos.) > > Basically, I like the XMLHttpRequest interface. I just think we need a > few simple things to capture common patterns in using XMLHttpRequest. > For me, this is captured in the "issue" function in my Request.js [1]. > > I know others like the loadURL signature. I don't see the need for > onComplete *and* onFailure (since there are many ways "failure" can > occur - including failure to parse XML which means your onComplete > function still needs "failure" checking code), but they are easy to > support as long as we keep loadURL around. > > Interested to hear any feedback. > > Tim > > [1] > http://trac.openlayers.org/browser/sandbox/tschaub/request/lib/OpenLayers/Request.js > >> <http://dev.openlayers.org/sandbox/tschaub/request/> >> <http://dev.openlayers.org/sandbox/tschaub/request/lib/OpenLayers/Request.js> >> <http://dev.openlayers.org/sandbox/tschaub/request/lib/OpenLayers/Request/XMLHttpRequest.js> >> >> That new code will support PUT and DELETE. Good Thing! >> >> Cheers, >> -- >> Eric
Tim, I like the proposal too, particularly that it makes it easier to set request headers. A few questions: Do the user and password parameters in example 8 indicate that you'd like to implement a layer of abstraction over HTTP authentication headers? I'd prefer to stay closer to bare HTTP. Proxying is a non-issue for me right now, but I wonder if OL really needs to know that it is making a proxy request. Couldn't it be transparent? Can we shout RFC 2616 at developers and have methods like Request.GET, POST, PUT, DELETE? Cheers, Sean _______________________________________________ Dev mailing list [email protected] http://openlayers.org/mailman/listinfo/dev
