(David's status) |
(→Minutes: fill in minutes) |
||
| (3 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
== Agenda == | == Agenda == | ||
* Selection of right Review Tool while being in a transition from MozReview to Phabricator | |||
** Should we continue using MozReview until EOL? Or immediately switch to Phabricator? Or use the old Splinter review? | |||
** MozReview will still allow creating new patches until August 26th | |||
** What's missing in Phabricator? | |||
*** Patch series aren't a supported feature yet, but extensions fill the gap for both hg and git | |||
**** https://bitbucket.org/kmaglione/hgext/src/default/phabricator.py | |||
**** https://github.com/mystor/phlay | |||
** Splinter is bad in many ways | |||
*** Updates of the patches remove former review status which require additional checks in bug comments to see details of a former review | |||
*** Issues as created during the review do not exist. It's all about comments ... | |||
*** ... which then even don't have to be replied to in Splinter, which makes following conversations very hard | |||
== Minutes == | == Minutes == | ||
=== Review tool === | |||
; ato | |||
: Problem with a rebase caused review status to be lost. Sorry about that. | |||
: [talks about issues with different review tools] | |||
: Lots of extra dependencies for those Phabricator extensions which are just a wrapper around arc/UI stuff. | |||
: As I understand it, Mark’s team is working on something built in to mach that will not require these dependencies but which instead will talk directly with Phabricator using its API. | |||
: Data loss for using Phabricator because it doesn't post back to Bugzilla. Could be potential problem if we move away from Phabricator in the future. | |||
; dburns | |||
: We should find a concensus for our team to make use of a single review system. | |||
; ato | |||
: Finding concensus can be hard when all the tools have problems. I think the best we can do is to suffer through a few weeks and then re-evaluate when support for our workflow is better with Phabricator. | |||
; whimboo | |||
: Splinter will also go away. | |||
; ato | |||
: Yes, but Splinter is essentially just comments on patchfiles. There’s no data loss because all comments are reflected in a single thread in Bugzilla. | |||
: Splinter is not great—mostly for UI reasons—but it’s simple and easy to understand. | |||
=== OKRs H2 === | |||
; dburns | |||
: Finishing up what we did in H1. | |||
: There's one OKR about using wpt.fyi data, helping other teams seeing where we have a low interop score. | |||
: Show Firefox vs. any browser and see how interoperable we are. | |||
; ato | |||
: This is so important, good effort. | |||
: But maybe difficult to get buy-in? | |||
; dburns | |||
: People want more web platform tests on mobile. | |||
; ato | |||
: People seem to think WPT tests and interop testing is important, but it's so far been hard to get engineering managers to insist on it. | |||
; dburns | |||
: Some of our harnesses have features that WPT doesn’t, and this is a problem (leak checks, ASAN) | |||
: Long list of reasons why developers are using mochitest instead of wpt, and we need to cut that list. | |||
[discussion about GeckoView] | |||
=== Interns next year === | |||
; ato | |||
: We have put in an intern application for next year for a “Gecko developer experience” project related to writing a W3C conforming WebDriver client and improving infrastructure. | |||
== Status updates == | == Status updates == | ||
| Line 9: | Line 61: | ||
* ato | * ato | ||
** Added debugging shorthand <code>Sleep</code> to sync module to pause current async task ([https://bugzilla.mozilla.org/show_bug.cgi?id=1473614 1473614]) | |||
** Removed <code>PREF_COMPLEX</code> from prefs module ([https://bugzilla.mozilla.org/show_bug.cgi?id=1480810 1480810]) | |||
** Dropped references to mozreview in documentation ([https://bugzilla.mozilla.org/show_bug.cgi?id=1479772 1479772], [https://bugzilla.mozilla.org/show_bug.cgi?id=1479774 1479774]) | |||
* Henrik | * Henrik | ||
** '''[wpt] Increased startup/shutdown timeout for Firefox to catch crashes caused by shutdown hangs (Bug 1456746)''' | |||
** '''[wdspec] Added remaining prompt behavior tests for existent user prompt tests (Bug 1479850)''' | |||
** '''[wdspec] About to move shared methods into __init__.py file for each command (Bug 1480337)''' | |||
** '''[wdspec] I will start working on Element Send Keys webdriver compatibility support soon''' | |||
** '''[geckodriver] Added geckodriver release mapping to Firefox and Selenium: https://github.com/mozilla/geckodriver/wiki/Geckodriver-Releases''' | |||
** '''[geckodriver] Continued converting code to Serde. Close to be able to build geckodriver (Bug 1396821)''' | |||
** [wdspec] Added tests for no browsing context (Bug 1478799) | |||
** [wdspec] Removed unnecessary createFile fixture for file upload tests (Bug 1479677) | |||
** [marionette] Fix for Execute Script if script body contains a comment in the last line (Bug 1337899) | |||
* David | * David | ||
Latest revision as of 14:28, 6 August 2018
Agenda
- Selection of right Review Tool while being in a transition from MozReview to Phabricator
- Should we continue using MozReview until EOL? Or immediately switch to Phabricator? Or use the old Splinter review?
- MozReview will still allow creating new patches until August 26th
- What's missing in Phabricator?
- Patch series aren't a supported feature yet, but extensions fill the gap for both hg and git
- Splinter is bad in many ways
- Updates of the patches remove former review status which require additional checks in bug comments to see details of a former review
- Issues as created during the review do not exist. It's all about comments ...
- ... which then even don't have to be replied to in Splinter, which makes following conversations very hard
Minutes
Review tool
- ato
- Problem with a rebase caused review status to be lost. Sorry about that.
- [talks about issues with different review tools]
- Lots of extra dependencies for those Phabricator extensions which are just a wrapper around arc/UI stuff.
- As I understand it, Mark’s team is working on something built in to mach that will not require these dependencies but which instead will talk directly with Phabricator using its API.
- Data loss for using Phabricator because it doesn't post back to Bugzilla. Could be potential problem if we move away from Phabricator in the future.
- dburns
- We should find a concensus for our team to make use of a single review system.
- ato
- Finding concensus can be hard when all the tools have problems. I think the best we can do is to suffer through a few weeks and then re-evaluate when support for our workflow is better with Phabricator.
- whimboo
- Splinter will also go away.
- ato
- Yes, but Splinter is essentially just comments on patchfiles. There’s no data loss because all comments are reflected in a single thread in Bugzilla.
- Splinter is not great—mostly for UI reasons—but it’s simple and easy to understand.
OKRs H2
- dburns
- Finishing up what we did in H1.
- There's one OKR about using wpt.fyi data, helping other teams seeing where we have a low interop score.
- Show Firefox vs. any browser and see how interoperable we are.
- ato
- This is so important, good effort.
- But maybe difficult to get buy-in?
- dburns
- People want more web platform tests on mobile.
- ato
- People seem to think WPT tests and interop testing is important, but it's so far been hard to get engineering managers to insist on it.
- dburns
- Some of our harnesses have features that WPT doesn’t, and this is a problem (leak checks, ASAN)
- Long list of reasons why developers are using mochitest instead of wpt, and we need to cut that list.
[discussion about GeckoView]
Interns next year
- ato
- We have put in an intern application for next year for a “Gecko developer experience” project related to writing a W3C conforming WebDriver client and improving infrastructure.
Status updates
(Spoken status updates in bold.)
- ato
- Henrik
- [wpt] Increased startup/shutdown timeout for Firefox to catch crashes caused by shutdown hangs (Bug 1456746)
- [wdspec] Added remaining prompt behavior tests for existent user prompt tests (Bug 1479850)
- [wdspec] About to move shared methods into __init__.py file for each command (Bug 1480337)
- [wdspec] I will start working on Element Send Keys webdriver compatibility support soon
- [geckodriver] Added geckodriver release mapping to Firefox and Selenium: https://github.com/mozilla/geckodriver/wiki/Geckodriver-Releases
- [geckodriver] Continued converting code to Serde. Close to be able to build geckodriver (Bug 1396821)
- [wdspec] Added tests for no browsing context (Bug 1478799)
- [wdspec] Removed unnecessary createFile fixture for file upload tests (Bug 1479677)
- [marionette] Fix for Execute Script if script body contains a comment in the last line (Bug 1337899)
- David
- OKRs for H2
PTO/travel (🏖)
- Monday 27 August is a bank holiday in the UK
- ato away Friday 10 August