Meeting minutes
TPAC Summary
Mike: More clarity on requirements from Apple. HID access is a non-starter.
The new learning is that they propose a spec that is built around user intent instead of letting the driver simulate any arbetrary keyboard command.
The BTT WG has a strong desire to see the intents sent to specific targets.
Meaning, the intent needs to be sent to a specific browser, and perhaps screen reader.
There should be more of an integration with webdriver so that we don't re-invent the management of sessions and windows.
James: These constraints could make the spec significantly less useful. Understand that there is some need to negotiate with stakeholder.s
I am specifically concerned about intent.
Micahel: +1
Michael: I do as well.
Matt: These constraints are motivated by security and platform-independence concerns.
Matt: there is strong support from several people in the browser testing tools WG for having an AT driver spec, including the chair.
Matt: there are also some people raising concerns and wondering if BTT is the right place for the spec. But they are not necessarily objecting.
Scope of AT Driver: is this now a protocol solely for the web?
Matt: There is a general concern that we shouldn
Matt: There is a general concern that we shouldn't publish a spec that applies beyond the web. Concerns raised regarded security, privacy, interoperability, etc.
James: The spec currently written does not contain anything that is specific to the target of testing. For example, how to download the AT, open the browser, and focus the window, etc. That's the job of the implementor.
James: for example, WebDriver is specific written to be a spec for browsers, but the tooling around it is the responsibility of the developer - not the w3c.
James: is the issue that we are doing this in the w3c and thus it needs to be explicitly tied to the web?
Matt: yes
Mike: the argument that this thing would ship on every machine is not necessarily true. for example, web driver doesn't ship with every version of Chrome.
(further discussion of similar concerns around WebDriver)
Matt: I haven't read the WebDriver spec, I should do that.
James: I understand the concerns, and this isn't a light decision.
James: we might end up in a place where one screen reader says 'we won't implement this without a sandbox' and another might say 'we can't implement a sandbox, and that would be too much work'.
Matt: I wonder if there is a way to allow for both ways and address all concerns.