1990203 - www.reddit.com - Copying a comment along with the user name copies the whole comment thread content

Status: RESOLVED FIXED

(Web Compatibility :: Site Reports, defect, P3)

Description

[Raul Bucata]

Environment:
Operating system: Windows 10
Firefox version: Firefox 145.0

Steps to reproduce:

1. Navigate to: https://www.reddit.com/r/roFrugal/comments/1nobyrl/donez_pizza_dodo/
2. Copy a comment that has replies and its user name
3. Paste anywhere and observe

Expected Behavior:
The title and comment made are copied

Actual Behavior:
The whole content of the comment thread is copied (usernames, replies, timestamp, etc)

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/178884

Comment 1

[Raul Bucata]

Attachment: Chr vs FF

Comment 2

[Raul Bucata]

Copying anything else other than this combination works. For example, copying just the comment pastes as expected.

Comment 3

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

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

Comment 4

[Sky Schubert [:skyschub]]

It doesn't do that for me on macOS, so maybe this is somehow oddly related to the Windows selection thing? But James could also see something odd on Linux, so maybe not.

Comment 5

[Kagami Rosylight [:saschanaz] (they/them)]

Hmm, can reproduce on Fedora.

Comment 6

[Kagami Rosylight [:saschanaz] (they/them)]

dom.shadowdom.selection_across_boundary.enabled=false fixes it.

Comment 7

[Kagami Rosylight [:saschanaz] (they/them)]

Each comment is composed of a custom element <shreddit-comment>, which has slot=commentMeta for nickname/timestamp and slot=comment for the comment body. Crossing the slot boundary seems to incorrectly copy the whole element (which includes subcomments as its children)

Comment 8

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

Set release status flags based on info from the regressing bug 1903870

:sefeng211, since you are the author of the regressor, bug 1903870, could you take a look?

For more information, please visit BugBot documentation.

Comment 9

[Kagami Rosylight [:saschanaz] (they/them)]

https://developer.mozilla.org/en-US/docs/Web/API/Web_components/Using_templates_and_slots#result causes the same issue, crossing the boundary between each details element unexpectedly causes full copy of both.

Comment 10

[Hsin-Yi Tsai (she/her)[:hsinyi]]

Hey Olli, adding this to your radar.

Comment 11

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

Set release status flags based on info from the regressing bug 1903870

Comment 12

[Edgar Chen [:edgar]]

This could also be related to the serializer, where the nsHTMLCopySelector "promotes" the selection range for serialization.

Comment 13

[Edgar Chen [:edgar]]

It is not about the range promotion, but the way we compute the flatten tree ancestors and offsets for serializing seems wrong, https://searchfox.org/firefox-main/rev/b67c6262ecbe7d4c9385a1bd335fd7dd1d3f887d/dom/serializers/nsDocumentEncoder.cpp#1415-1420. I think the root cause is the same as bug 1994763.

Comment 14

[Edgar Chen [:edgar]]

I have verified on today's nightly (2026-01-30), it is fixed now.