1884763 - Some elements with `translate3d` won't render

Status: UNCONFIRMED

(Core :: Web Painting, defect)

Description

[juraj.masiar]

Attachment: left_chrome_right_firefox.png

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:124.0) Gecko/20100101 Firefox/124.0

Steps to reproduce:

I'm trying to create a 3D effect for my app but there is a strange bug that causes some DIV elements not being rendered or not being clickable.
It works fine in Chromium, only Firefox is affected.

I've tried to recreate it in this sandbox, but it doesn't reproduce every time, so if you see all 16 squares, try to reload the page until you see 14 of them:
CodeSandbox link

Actual results:

Only 14 squares is rendered, the two on the left are missing.

Expected results:

All 16 squares should be visible.
See also attached image.

Comment 1

[Emilio Cobos Álvarez (:emilio)]

I haven't been able to reproduce this so far. Can you attach your about:support information? It'd also be useful to check if this repros in troubleshooting mode or what not.

Comment 2

[juraj.masiar]

Attachment: about:support

Did you tried ESR or Release or Developer Edition? I can reproduce it in these, ether on first load or 100% on the second load (press F5). I can reproduce this even in Android.
But somehow I can't reproduce it in Nightly, but my app I'm developing fails also in Nightly, it's for sure broken there too.

```json

Comment 3

[juraj.masiar]

Attachment: android, Firefox Beta 124.0b9

Comment 4

[Emilio Cobos Álvarez (:emilio)]

Can try, thanks for the extra context!

Comment 5

[Timothy Nikkel (:tnikkel)]

It seems easy for me to reproduce in beta builds, but I haven't been able to reproduce in nightly.

Comment 6

[juraj.masiar]

From what I've seen, the bug reacts to two things:

1. using higher negative "tz" value (sending the squares further back) is what's trigger the "invisibility". When I use only about -200px, they will be drawn, but they won't be clickable
2. it won't trigger without the blank "img" node in the first square. So maybe adding more images will make it easier to reproduce (in my app there is many images and the bug is always present, even in Nightly).

Comment 7

[Emilio Cobos Álvarez (:emilio)]

Attachment: Standalone test-case from comment 0.

I can repro on beta. I can also repro on Nightly at some different zoom levels (e.g., for me it repros at 130% with this test-case).

Comment 8

[Timothy Nikkel (:tnikkel)]

Changing zoom helps.

I got this regression range (double checked it)

https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=51f2535c797495f1f4e864072c2449b7c28669de&tochange=96a6ea5ea3468b4c9e20ff8d9795a7ef136213a9

Comment 9

[Emilio Cobos Álvarez (:emilio)]

According to mozregression this is an ancient regression from https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=bb8d6034f5f2&tochange=f5947d58ab02

I'm on Linux, and the plausibly-related change I found is bug 1119117. Yet there's no opacity at all in the test-case so I'm confused...

Comment 10

[Emilio Cobos Álvarez (:emilio)]

(comment 9 mid-aired with comment 8)

Comment 11

[Emilio Cobos Álvarez (:emilio)]

I don't understand how the regression ranges from comment 9 and comment 8 can be so different...

Comment 12

[Emilio Cobos Álvarez (:emilio)]

Attachment: More reduced test-case.

Comment 13

[Emilio Cobos Álvarez (:emilio)]

I can definitely repro the issue with FF50: MOZ_DISABLE_CONTENT_SANDBOX=1 mozregression --launch 50 -a t.html

Comment 14

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

The severity field is not set for this bug.
:tnikkel, could you have a look please?

For more information, please visit BugBot documentation.