WebDriver/RemoteProtocol/Meetings/2020/03/25: Difference between revisions

Jump to navigation Jump to search
add notes
(→‎Agenda: added agenda item for upstream syncs)
(add notes)
Line 8: Line 8:
** How often should we do upstream/downstream syncs in the future?
** How often should we do upstream/downstream syncs in the future?
** Fixing issues with tests locally or in upstream repository first (eg. meta key)?
** Fixing issues with tests locally or in upstream repository first (eg. meta key)?
* [harald] Dashboards
* [harald] Explorations


== Roster ==
== Roster ==
Line 13: Line 15:
; Present
; Present


: mmucci, maja, harald, henrik, jgraham, joeleinbinder (Microsoft)


== Minutes ==
== Minutes ==


=== Status update ===
mmucci - Status update: on track, slightly ahead of schedule
=== Process of stay in sync with the Puppeteer upstream repository ===
* [maja] The next sync might be a pain but not that bad; subsequent syncs won't be a problem
* [james] Lets make changes in m-c and sync every once in a while
* henrik: easier to do in m-c only
* Action - Maja will run a sync soon.
=== Dashboard ===
* Harald: https://github.com/puppeteer/puppeteer/issues/5515#issuecomment-600783317
* Migrate ispuppeteerready dashboard into puppeteer org
* Can Joel help?
* James - we don't have the same type of information as the old dashboard so it's not super important to transfer ownership
* Harald - we want to make sure old one eventually points to the up-to-date dashboard
* Joel - old dashboard was based on the coverage tests
* Joel will ping Andrey
* Harald: report based on data we have now or wait until we have richer data?
* Conclusion: go ahead with our current data, don't block on ownership of previous dashboard
=== Explorations ===
* Harald: Relationship of Remote Protocol to Marionette, RDP; convergence?
* James: expect to have an upgrade path from WebDriver to bidi protocol, so bidi has same featureset as WebDriver plus more. Similarly, Medium/long term goal is that Marionette goes away and its featureset is build on top of new bidi protocol. However, some Marionette features are Firefox-specific (e.g. access to browser chrome). In the short term ktlo for Marionette + Fission changes. RDP unaffected.
** Maja: geckodriver is venue for transition from Marionette to bidi remote protocol; transition invisible to WebDriver users
** James: internal users of Marionette would have to transition to bidi remote protocol
** Harald: super power in Remote Protocl to access browser chrome for example?
** Henrik: some architectural issues anticipated with that. In geckodriver right now we offer the ability to switch to chrome context but as far as I know only Dave Hunt's FoxPuppet client is using chrome context via geckodriver
** Maja: as WebDriver, Remote Protocol will have to allow for browser-specific extensions (such as browser chrome access)
** James: devtools features should not be in scope for Remote Protocol proposal
** Harald: browser vendors want freedom to break their devtools apis
** Joel: my understanding of CDP is the most stable in the devtools APIs (because shared with node) and least stable in the automation APIs
** James: CDP debugging APIs less stable
** Harald: console and debugger space
* Harald: delta between remote protocol and cdp?
** Joel: flexibility in CDP, not being locked into architecture; like WebKit people put everything in automation domain
* Harald: would it make sense to switch RDP to bidi Remote Protocol? Convergence of remote protocols
** Maja: Some parts of CDP can be carried over into remote protocol with only minor changes
** James: Depends on if the APIs can be matched by all browsers
** James: Overlap of automation and devtools needs, like getting network requests
=== Playwright updates from Joel ===
* WebKit parts looking good; coming back to Firefox work and clean it up
* Joel will try to set up a dashboard for Playwright against Firefox Nightly
* 1.0 of Playwright
** Wait till end of March, stabilize. Release mid-April


== Progress ==
== Progress ==
Confirmed users
378

edits

Navigation menu