Notes

• This behavior is seen using amazon and ddg search engines so far. It also reproduces when setting DDG or amazon as default search engines. Please see the screenshoot for more details.

Found in

• Nightly 111.0a1

Affected versions

• Nightly 111.0a1
• Firefox 110.0

Tested platforms

• macOS 12
• Windows 10
• Ubuntu 22

Affected platforms

• macOS 12
• Windows 10
• Ubuntu 22

Unaffected platforms

• N/A

Steps to reproduce

1. Launch Firefox.
2. Focus the address bar and select the DDG or the Amazon one-off search engines.
3. Type a string in the Address Bar and submit the search.

Expected result

• One recorded entry is present in the History section for the searched string.

Actual result

• Two recorded entries are present in the History section for the searched string.

Regression range

• First bad: 2022-02-19
• Last good: 2022-02-18
• Pushlog: https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=2353e006b0d70e1760eb02917eb959bd71ce7eed&tochange=797bfebd963c4716c4808300e8960b48bff101e3
• Potentially regressed by: bug 1754426

(In reply to Ardelean Oana from comment #0)

Regression range

• First bad: 2022-02-19
• Last good: 2022-02-18
• Pushlog: https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=2353e006b0d70e1760eb02917eb959bd71ce7eed&tochange=797bfebd963c4716c4808300e8960b48bff101e3
• Potentially regressed by: bug 1754426

I'm dubious of this. The bug referenced affected code that was disabled in m-c at the time and doesn't have impact on the history tables themselves

Set release status flags based on info from the regressing bug 1754426

I don't think this is a regression, it's likely they just changed something on the website, or it's some kind of user agent version sniffing.
I see the same behavior on Chrome, and it's likely related to bug 753264. They are very likely using replaceState to add parameters to the url, and we never investigated what global history should do with replaceState, since there's no spec around global history.
See https://bugzilla.mozilla.org/show_bug.cgi?id=753264#c3 and following.

I agree with the final call there that this is a problem that should be handled at the UI level, the global history behavior is not wrong per se, and changing it may introduce dataloss. On the other side the ui can "easily" deduplicate entries, and if it's missing information to do so, global history can be improved to store and provide such information.