[aik099]

Since ChromeDriver 134 a click that triggers navigation (e.g. form submission) sometimes returns before the new page is loaded, or the navigation is cancelled entirely: https://issues.chromium.org/issues/402796660

As a result, Form tests in Chrome-based builds (chrome, chromium, edge) with Selenium 4 fail randomly (1-3 per job, different each run) because they read content of the previous page, e.g. Submit empty textarea, Advanced form second submit, Basic get form, Html 5 form radio attribute. See minkphp/driver-testsuite#110 for the discussion.

Adding the --disable-features=DeferRendererTasksAfterInput browser argument makes the ChromeDriver wait for navigation again (found in teamcapybara/capybara#2800). The argument is added to the default Chrome and Edge capabilities, so users get the fix too.

Locally on selenium/standalone-chromium:4, Form tests passed in 1 of 6 runs without the argument and in 6 of 6 runs with it.

Note: as before, a user-provided goog:chromeOptions (or ms:edgeOptions) capability replaces the default one entirely, so in that case the argument needs to be added by the user.

This PR doesn't include the fix from #67, so WindowTest hangs will still be present here.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 88.86%. Comparing base (acd45d2) to head (4d07cf6).

Additional details and impacted files

@@             Coverage Diff              @@
##               main      #68      +/-   ##
============================================
- Coverage     89.06%   88.86%   -0.21%     
  Complexity      197      197              
============================================
  Files             1        1              
  Lines           494      494              
============================================
- Hits            440      439       -1     
- Misses           54       55       +1     

Flag	Coverage Δ
chrome	88.25% <ø> (-0.21%)	⬇️
chromium	87.85% <ø> (ø)
edge	88.25% <ø> (ø)
firefox	88.05% <ø> (ø)
selenium-2.53.1	87.04% <ø> (ø)
selenium-3	88.05% <ø> (ø)
selenium-4	88.46% <ø> (-0.21%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[aik099]

Comparison of the reference main build (run) vs this PR (run):

Matrix entries	main	This PR	Result
Selenium 4: chrome (PHP 7.4-8.3), chromium, edge	❌ random Form failures in 5 of 7 jobs (Submit empty textarea, Multi input, Basic get form, Advanced form, Html 5 types, etc.) + WindowTest hangs	❌ WindowTest hangs only, no Form failures in any of 7 jobs	Form failures fixed
Selenium 4: firefox	❌ HoverTest + WindowTest hangs	❌ same	not affected (not Chromium-based)
Selenium 3: chrome, firefox	❌ WindowTest hangs	❌ WindowTest hangs (firefox only)	not affected
Selenium 2.53.1: chrome, firefox	✅ pass	✅ pass	no change
Static analysis	❌ unmatched ignoreErrors patterns	❌ same	unrelated

The WindowTest hangs are fixed in #67 and the Firefox HoverTest failures will be addressed separately.

For extra context: in the 4 CI runs of #67 (which doesn't include this fix), Form failures happened in 4-7 of the 7 Chrome-based jobs on every run.

[aik099]

Rebased on main after #67: all Chrome-based jobs (chrome, chromium, edge) are now green (run). The only remaining failures are Firefox + Selenium 4 HoverTest (#69) and static analysis.

[stof]

Looks good to me.
We should probably do the same in Selenium2Driver btw.