1948861 - www.instagram.com - Scrolls to the top of the page when trying to share reels (reproduces randomly)

Status: NEW

(Web Compatibility :: Site Reports, defect, P2)

Description

[Calin Tanase]

Environment:
Operating system: Windows 10
Firefox version: Firefox 137.0/135

Preconditions:

• Must be logged in

Steps to reproduce:

1. Go to https://www.instagram.com/reels
2. Scroll through the reels a little.
3. Click on the share/send button located on the right of a reel.
4. Close the share window.
5. Observe the behavior.

Expected Behavior:
Displays the same reel you were trying to share.

Actual Behavior:
Scrolls to the top, it display the first reel.

Notes:

• Reproduces regardless of the status of ETP
• Reproduces in firefox-nightly, and firefox-release
• Does not reproduce in chrome

Created from https://github.com/webcompat/web-bugs/issues/148283

Comment 1

[Calin Tanase]

Attachment: IG FF vs Chrome.mp4

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

Since nightly and release are affected, beta will likely be affected too.
For more information, please visit BugBot documentation.

Comment 3

[Dennis Schubert [:denschub]]

I don't know if they're using scroll snapping, but they might?

Comment 4

[Randell Jesup [:jesup] (needinfo me)]

hiro: Is this related to bug 1948863?

Comment 5

[Hiroyuki Ikezoe (:hiro)]

Attachment: WIP: Bug 1948861 - Do not clobber restoring the scroll position on re-snap.

Comment 6

[Hiroyuki Ikezoe (:hiro)]

D239677 fixes this bug, and I think it's the right approach. But I haven't been able to write any test cases for this bug yet.

(In reply to Randell Jesup [:jesup] (needinfo me) from comment #4)

hiro: Is this related to bug 1948863?

I would say yes, it's related. Both this bug and bug 1948863 happen with scroll snap. But given that alice has already subscribed this bug, but hasn't yet commented here, so I presume the underlying issue is different.

That's being said, I am not 100% sure because bug 1948863 is hard to reproduce on my end, D239677 might fix bug 1948863 altogether.

Comment 7

[Hiroyuki Ikezoe (:hiro)]

Attachment: WIP: Bug 1948861 - Clear mSnapTargets just before trying to resnap instead of inside ScrollContainerFrame::GetSnapPointForResnap.

Though there's no problem even if we clear mSnapTargets in
ScrollContainerFrame::GetSnapPointForResnap, the main purpose of this change
is to avoid multiple call sites of ScrollSnapUtils::GetSnapPointForResnap
so that in the next change we only need to tweak the single call site.

Comment 8

[Daniel Holbert [:dholbert]]

[moving to diagnosis-team:apz since hiro's looking into this and it seems to be APZ/scrolling sort of issue.]

Comment 9

[Hiroyuki Ikezoe (:hiro)]

Attachment: A test case

1 sec. after loading the test case, you will see "scrollPosition: XX" there, if things work properly, the number should be 200, but on Firefox it's 0.

Comment 10

[Hiroyuki Ikezoe (:hiro)]

Daniel noticed me that I should have opened a platform bug and attached the patches there.

Comment 11

[Phabricator Automation]

Comment on attachment 9468527 [details]
WIP: Bug 1948861 - Clear mSnapTargets just before trying to resnap instead of inside ScrollContainerFrame::GetSnapPointForResnap.

Revision D239685 was moved to bug 1950560. Setting attachment 9468527 [details] to obsolete.

Comment 12

[Phabricator Automation]

Comment on attachment 9468514 [details]
WIP: Bug 1948861 - Do not clobber restoring the scroll position on re-snap.

Revision D239677 was moved to bug 1950560. Setting attachment 9468514 [details] to obsolete.

Comment 13

[Hiroyuki Ikezoe (:hiro)]

For posterity;

Note that what I diagnosed is that in nsVideoFrame::ReflowFinished we dispatch a resizevideocontrols event and the resizevideocontrols event triggers re-snapping before restoring the scroll position via this.updateReflowedDimensions() call.

Comment 14

[Hiroyuki Ikezoe (:hiro)]

Geez, I intended to leave comment 13 in bug 1950560, not here.

Comment 15

[Botond Ballo [:botond]]

Replacing "webcompat:needs-diagnosis" with "webcompat:platform-bug" since this has been diagnosed as being caused by a platform bug tracked in bug 1950560.