New issue
Advanced search Search tips

Issue 594011 link

Starred by 1 user
Status: Duplicate
Merged: issue 595634
Owner:
yangguo@chromium.org
Closed: Mar 2016
Cc:
jochen@chromium.org
mvstan...@chromium.org
yangguo@chromium.org
hablich@chromium.org
bokan@chromium.org
danno@chromium.org
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug



Sign in to add a comment

Pasting issue in XenForo Reply Box

Reported by sush...@gmail.com, Mar 11 2016 
UserAgent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/51.0.2673.0 Safari/537.36

Example URL:
Any XenForo forum

Steps to reproduce the problem:
1. Login to any XenForo forum.
2. Open a thread.
3. Go to the bottom to the Quick Reply box. Paste some text from the clipboard. 

What is the expected behavior?
Text is pasted instantly without any delay. 

What went wrong?
Text is pasted after slight delay. This happens only on the first paste after a page load. If you paste again, it pastes instantly. If you refresh the page, there is again a slight delay. 

Does it occur on multiple sites: Yes

Is it a problem with a plugin? No 

Did this work before? Yes There are no such issues in the Stable build right now.

Does this work in other browsers? Yes 

Chrome version: 51.0.2673.0  Channel: canary
OS Version: 10.0
Flash Version: Shockwave Flash 21.0 r0

I can provide a test link on a XenForo powered forum that does not require login. Posting it publicly here could cause spam post. 

This issue has been in Canary for a long time now. Recently entered Dev build. I have not checked the beta edition as I moved to stable. Stable builds do not show this issue.

Comment 1 Deleted

Comment 2 by jochen@chromium.org, Mar 14 2016 

there's a huge jquery execution taking place in the middle.

Are you saying that during your bisect, this block is considerably shorter or goes away entirely? Otherwise, I don't see how this was a v8 issue

Comment 3 by bokan@chromium.org, Mar 14 2016 

Right, the paste is near instant prior to that commit.

Comment 4 by jochen@chromium.org, Mar 15 2016 

my question was, when you take the trace on a "good" revision, what does it look like then?

Comment 5 by jochen@chromium.org, Mar 15 2016

Cc: jochen@chromium.org

Comment 6 by bokan@chromium.org, Mar 15 2016 

Ah, sorry. I attached a trace (pid 21707) from a good rev. It has the same V8 execution but it's only ~100ms.
trace_paste_delay_good.json.gz
3.3 MB Download

Comment 7 by jochen@chromium.org, Mar 16 2016

Cc: mvstan...@chromium.org
Owner: yangguo@chromium.org
Yang, I got the same regression range when bisecting. It looks like the V8 roll introduced some regex changes, maybe that is what hits jquery so badly?

On my local machine, we slow down jquery by 5x

Comment 8 by jochen@chromium.org, Mar 16 2016

Cc: danno@chromium.org

Comment 9 by yangguo@chromium.org, Mar 16 2016 

Since you are already bisecting, could you go the extra mile and bisect into the V8 roll?

The regexp change is behind a flag, so I'm a bit sceptical.

Comment 10 by hablich@chromium.org, Mar 16 2016

Cc: yangguo@chromium.org
Owner: jochen@chromium.org
Status: Assigned (was: Untriaged)
Strange that our own jQuery tests don't show that: https://chromeperf.appspot.com/report?sid=c870fc5daefc533790c9c0ba5c884e45ee8ea30974bf74eaaa62b179e0d50934&start_rev=368716&end_rev=371716

Comment 11 by jochen@chromium.org, Mar 16 2016

Owner: yangguo@chromium.org
bisected this down to https://chromium.googlesource.com/v8/v8/+/e709aa24c0c17abf684972fbb9e887731b20fd41

Comment 12 by jochen@chromium.org, Mar 17 2016

Mergedinto: 595634
Status: Duplicate (was: Assigned)
Merging to  issue 595634  which has simpler repro steps
►

Sign in to add a comment