- From: Robin Berjon <robin@w3.org>
- Date: Fri, 21 Mar 2014 11:53:39 +0100
- To: Odin Hørthe Omdal <odinho@opera.com>, public-test-infra@w3.org
On 21/03/2014 11:44 , Odin Hørthe Omdal wrote: > On Thu, Mar 20, 2014, at 18:14, James Graham wrote: >> wptserve doesn't really support the kind of wildcard redirect you are >> looking for. Since it's not a feature required to test the web platform >> I am reluctant to add it unless we are really sure it's worth the cost. > > Can wpt be proxied through nginx which has the features? > > Though proxying through will of course make certain tests fail. So it's > a bad idea anyway. That's what we wanted to go away from. Indeed. Please, let's not go back to that. > Is there another (cleaner) way to have wptserve defer to some other > authorithy on cases it doesn't know how to handle? (a fall-through?). I agree with James that it's probably a bad idea to introduce mod_rewrite into wptserve. It's supposed to stay simple so that it can reliably be tested with. That said, maybe we could introduce a feature that would keep things relatively simple: a 404 handler. Basically, the JSON configuration can have a field pointing to file that is called whenever wptserve hits a 404. That file is called exactly like the server-side bits we already have. People who then want to redirect stuff can then hack that in Python. Having a single possible handler and forcing people to use Python should make for a suitably high barrier to usage while keeping implementation relatively simple. -- Robin Berjon - http://berjon.com/ - @robinberjon
Received on Friday, 21 March 2014 10:53:52 UTC