Re: Apparent WebDriver meeting?

Thanks for that great summary, Philip.There's definitely no intention of
keeping this meeting "hidden" or secret. Anyone who's interested in talking
about BiDi is welcome to join.

As for logistics, I suggest we simply invite
public-browser-tools-testing@w3.org to any future meetings, and start
taking public minutes. WDYT?

On Tue, Apr 21, 2020 at 1:30 PM Philip Jägenstedt <foolip@google.com> wrote:

> Hi David,
>
> A bunch of people indeed meet yesterday to discuss the explainer
> <https://github.com/w3c/webdriver/blob/master/webdriver-bidi/webdriver.md> and
> some of the BiDi issues <https://github.com/w3c/webdriver/labels/BiDi>
> filed by James. This grew organically out of a first Google/Microsoft
> meeting to gauge the level of interest about a month ago. Since it's such
> early days for this workstream, I don't think any of us have considered
> whether to use existing WG meetings, public minutes, etc., we just met to
> see what would come of it.
>
> Trying to summarize some of the discussion:
>
>    - Specify how to enable an event stream
>    <https://github.com/w3c/webdriver/issues/1501> - The basic problem is
>    the full stream of events might be too much, so you'd need some way to
>    enable a subset of all possible events. The next problem is that there are
>    some events from targets you don't know if/when they'll appear, like
>    iframes and service workers, that you want to subscribe to if they appear.
>    We thought that some per-session mechanism to enable events per domain and
>    possibly per target type might work.
>    - Update goals in the explainer document
>    <https://github.com/w3c/webdriver/issues/1499> - I think we ended up
>    concurring with what James wrote on devtools protocols. While being able to
>    use BiDi together with devtools protocols would probably help a transition,
>    it's no requirement that BiDi is layered on top of a devtools protocol, and
>    for example how to get from identifiers in one protocol to the other, when
>    both are supported, would perhaps be as extra properties or "convert this
>    to that" commands in one of the protocols.
>    - Specify a low-level transport format
>    <https://github.com/w3c/webdriver/issues/1498> - It sounds like
>    JSON-RPC isn't exactly what we need, but there isn't anything else that is.
>    Basing something on JSON-RPC but writing a stringent test suite for it
>    sounded reasonable. It's not clear if transmitting binary as base64 will be
>    enough of a problem to avoid JSON-RPC, I got the sense that most
>    implementers don't think so.
>
> @Mathias Bynens <mths@google.com> or others on this list might be able to
> add more detail, and in any case it'd be a good idea to get this into the
> issues rather than notes.
>
> I've not seen any suggestion to spin up a CG for this, but enough people
> found it useful enough to repeat it on a monthly basis. I think it would
> make sense to have the BiDi meeting be open to any WG members and to take
> minutes in public. What are the logistics for doing that?
>
> Philip
>
> On Tue, Apr 21, 2020 at 10:56 AM David Burns <
> david.burns@theautomatedtester.co.uk> wrote:
>
>> Hi All,
>>
>> I am just sharing this as there appears to be hidden meetings about the
>> WebDriver BiDi work that does not include anyone from the Selenium group,
>> the editors of the webdriver specification and probably others (who knows
>> as it was a closed meeting).
>>
>> Has there been some agreement to move this out to a community group
>> instead of using the working group?
>>
>> David
>>
>