- From: James Graham <james@hoppipolla.co.uk>
- Date: Tue, 16 Jun 2015 18:10:21 +0100
- To: Rick Byers <rbyers@chromium.org>, Doug Schepers <schepers@w3.org>
- CC: public-browser-tools-testing@w3.org, Pointer Events WG <public-pointer-events@w3.org>, Philippe Le Hegaret <plh@w3.org>, Mike Smith <mike@w3.org>, Scott González <scott.gonzalez@gmail.com>, Art Barstow <art.barstow@gmail.com>
On 16/06/15 17:46, Rick Byers wrote: > Thanks for kicking this off Doug, > > To me the main issue to discuss is whether / how the WebDriver API > should be extended to support rich input support (it's really too > limited at the moment to be that useful for scenarios like pointer events). > > I've started a blink-specific thread to discuss using WebDriver in our > tests here > <https://groups.google.com/a/chromium.org/forum/#%21topic/blink-dev/0lXHWPF4lD4>. It would be nice if we could avoid fragmenting the discussion so early. So: * Using WebDriver for doing automated testing of (the content area of) browsers has always been part of the plan going forward. * How to do this in a really nice way, browser-independent, isn't clear. All the existing W3C test infrastructure is written in python, so adding a node dependency would be unwelcome. Making authors write tests spanning multiple files is also additional overhead compared to the state of the art in browser-internal test frameworks. The best idea so far is to attempt an in-browser webdriver implementation that will allow calling out (via XHR or WS, or something) to an external server implementing a WebDriver client, that will inject the WebDriver commands and return the results. I'm sure this can work, but I am a little scared of the complexity. * The actions API is being specified to provide lower-level input support. You can see this in the current spec, but the actual text needs a bunch of work (more than any other part of the spec).
Received on Tuesday, 16 June 2015 17:10:56 UTC