On May 26, 2011, at 2:08 AM, Francisco Costa wrote: > > any update on this?
No, I just got back from a trip and haven't looked at it. Can we discuss what behavior we want (and can get)? The easiest option: a flag to forbid a cross-domain URL call. URL would raise an exception if you try to do that. (Actually, the rewrite logic would do that.) What's not so easy is to add a domain to the generated URL. Right now, the rewrite logic never adds a domain (and there's no mechanism to do so), and in fact doesn't even see the host & port arguments that URL() is called with. This is in large part because of the historical divide between URL() itself and the routing logic. Note also that it's not obvious which domain to associate with an outgoing URL, since multiple domains can map to a single app (more generally multiple host:port combinations), so we'd have to tag one domain as the "master" domain for outgoing URLs in a given app (maybe by convention the first one specified?). And we'd have to know whether to make it http or https. > > On May 24, 2:36 am, Jonathan Lundell <jlund...@pobox.com> wrote: >> On May 23, 2011, at 8:06 PM, Bruno Rocha wrote: >> >>> I guess Francisco wants to explain a different behaviour. >> >>> Having 2 or more apps on the same web2py instance, web2py should prevent >>> the app to be requested by other domain than its own domain. >> >>> www.app1.com-> /app1 #allowed >>> www.app2.com-> /app2 #allowed >> >>> www.app1.com/app2-> /app2 # denied >>> www.app2.com/app1-> app1 # denied >> >>> I agree with him, sometimes search robots finds apps under different >>> domains... >> >> We can do that, I think, easily enough. But I'm not so sure about >> automatically generating an absolute URL in UIRL() for a "foreign" >> domain-app combination. Maybe... >> >> >> >> >> >> >> >> >> >>> -- >>> Bruno Rocha >>> [ About me:http://zerp.ly/rochacbruno] >> >>> On Mon, May 23, 2011 at 9:40 PM, Jonathan Lundell <jlund...@pobox.com> >>> wrote: >>> On May 23, 2011, at 6:10 PM, Francisco Costa wrote: >> >>>> The URL calling is fine.. the problem is how web2py manages different >>>> apps from different domains/subdomains >> >>>> I really believe that they should be improved! >> >>> web2py (URL) isn't generating a domain *at all* in the URL unless you >>> explicitly specify one; the domain is normally provided by the browser, and >>> that's how it should be, at least by default. If you want to override the >>> domain the browser is using, you need to do it explicitly. >> >>>> On May 23, 11:02 pm, Jonathan Lundell <jlund...@pobox.com> wrote: >>>>> What's your URL() call for the example below? Have you looked at the >>>>> HTML? I wouldn't expect to see a domain there at all. >> >>>>> On May 23, 2011, at 12:00 PM, Francisco Costa <m...@franciscocosta.com> >>>>> wrote: >> >>>>>> Hello, >>>>>> I have a main app and a blog app >> >>>>>> My routes.py are like this: >>>>>> routers = dict( >>>>>> BASE = dict( >>>>>> default_application = 'main', >>>>>> default_controller = 'default', >>>>>> default_function = 'index', >>>>>> domains = { >>>>>> "domain.com" : "main", >>>>>> "blog.domain.com" : "blog", >>>>>> } >>>>>> ) >>>>>> ) >> >>>>>> While I'm browsing through the subdomain "blog.domain.com" I have some >>>>>> links that point to the main app. >>>>>> Those links are generated like this: >>>>>> http://blog.domain.com/main/test >> >>>>>> If I click on it it works, but I would like the link to be generated >>>>>> as >>>>>> http://domain.com/main/test >> >>>>>> This is bad because 2 different domains urls shouldn't target the same >>>>>> page and it also duplicates data which is bad for SEO >> >>>>>> Any help?
