- From: James Graham <jgraham@opera.com>
- Date: Thu, 26 Jul 2012 10:42:04 +0200
- To: public-coremob@w3.org
On 07/21/2012 09:46 AM, Jo Rabin wrote: > On 20 Jul 2012, at 22:58, SULLIVAN, BRYAN L wrote: > >> OK, whether there is a document or not, my goal for this CG is to >> help promote tools that help ensure a consistent feature set is >> available across as many browsers as possible. In the case of Level >> 0, I suggest we do the following: >> >> 1) Identify what we consider to be the expected set of features, >> via a document and test suite or a test suite alone > > My view is that a document is required and that a start on a test > suite is highly desirable. The features identified in the document > are chosen based on a set of mobile Web application use cases. To repeat a previous point, we should not, ourselves, be looking to develop a testsuite. If people have the resources to develop tests (and that is much harder to find than agreement that having tests is a nice idea) we should be contributing those tests to the relevant W3C working group. Having multiple testing efforts without coordination doesn't help anyone. >> 2) To help fill the cracks as we find them, identify reliable >> polyfill libraries that bring the deficient browsers up to par > > That sounds interesting. Bearing in mind what I think we agreed as > the test methodology - that would mean that we are testing > device/browser(version)/{library(version)}* combinations. What I mean > by that is that each library can occur in multiple versions, in > combination with multiple other libraries each at varying versions in > the context of multiple browsers each at varying versions for any > particular device. Sounds a bit scary, tbh. I think people are overestimating the level at which one starts to find interoperability problems. Very simple things like page loading, history navigation, form submission and document parsing have an alarming number of inconsistencies. Actually parsing is a really good example of a success story here; a year or two ago browsers all shipped with incompatible HTML parsers. The advent of a very detailed specification and a high quality testsuite (thousands of tests) has allowed everyone to converge on almost identical behaviour. A true "Level 0" effort would focus on the fundamentals of the web browser, much of which is below the level that can be easily "polyfilled".
Received on Thursday, 26 July 2012 08:42:35 UTC