Scott Babcock

The referenced PR doesn't actually resolve the issues raised by this RFC. It actually makes the situation slightly worse by adding another entry to our list of special-case extension capability...

Another set of changes has been applied to **DefaultSlotMatcher** and **RelaySessionFactory** that introduces additional Appium-specific implementation. I think this is the wrong direction, because this code should be vendor-agnostic. https://github.com/SeleniumHQ/selenium/commit/4cacd4a05b9dea9185fec25e6442e07f5fecf0ce

@joerg1985 As I read the W3C specification, the slot matcher should not be interpreting custom capabilities in any way. These should be entirely ignored and passed on without examination. Unfortunately,...

@diemol The main issues with the existing implementation are that it treats the "non-special" extension capabilities different than it does the "special" ones, and it alters the requested client capabilities...

@VietND96 Capabilities with "simple" values (Boolean/Number/String) will be considered for slot matching, even if their capability names specify a prefix. Capabilities with "complex" values (Array/Object) will be ignored for slot...

@VietND96 I'm trying to understand why the comparison in the test you added should return `true`. Not only are the values for `moz:debuggerAddress` not the same; the values are of...

@diemol Changing the behavior to only ignore extension capabilities whose names contain "options" would produce the same behavior that is causing the test @VietND96 added to fail. The values of...

@diemol I understand that we currently ignore all "special" extension capabilities, but there appears to be no real-world scenarios of "special" extension capabilities with "simple" values in which this behavior...

@diemol Why is it desirable to support the scenario enforced by the test that @VietND96 added? We match nodes even when capabilities like `moz:debuggerAddress` are entirely different between requested and...

@diemol I force-push after rebasing to keep the commit history clean. My attempt to rebase after @VietND96 added his unit test went horribly wrong and I had to check out...