1881801 - Twitter iframes slow to appear

Status: RESOLVED FIXED

(Core :: Networking, defect, P2)

Description

[Nojevah]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:123.0) Gecko/20100101 Firefox/123.0

Steps to reproduce:

Go to this link: https://www.lequipe.fr/Basket/Actualites/Rising-stars-challenge-victor-wembanyama-prive-de-finale/1449281

Actual results:

The twitter frames are slow to display if you scroll to the bottom.

Expected results:

If you compare with Chromium browser, you'll see they are quicker to appear.

More explanation:
All my tests have been done with a new profile, User Agent has my information.
Unfortunately, there are cookie notices ads etc. The ideal would be to block them with ublock (since behavior is the same with or without it) or a DNS Proxy. I alos have a good optical fiber connection.

On this link, there are 3 Twitter iframes.
First, I reload the page, then wait, no scrolling. If the network console is opened, you can clear it.
Then I scroll to the bottom of the article.
=> The 3 Twitter iframes need more than 10 seconds to all appear (the first one can be quicker).
Looking at network tab, there's nothing strange (everything way under 1 second).

I guess it's a problem with the website or Twitter, but I feel the need to report the issue, because Chromium seems to handle this better.

Comment 1

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

The Bugbug bot thinks this bug should belong to the 'Core::Networking' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 2

[Valentin Gosu [:valentin] (he/him) {{ AFK in July }}]

Thank you for the report. It's definitely an interesting test case, and I can confirm the iframes do load much later in Firefox.

If let me know if you have time to dig into this.
I've uploaded a Firefox profile for my run: https://share.firefox.dev/3OVLOYj
The page is very resource heavy, so if we don't see any things we might be doing wrong, it could be a prioritization issue, in which case we should look at what Chrome is doing better. I've captured a perf profile in Chrome too. link here

Comment 5

[Nojevah]

• In my first post: "All my tests have been done with a new profile"
• It has been reproduced by Valentin from Mozilla.

It doesn't need support, it needs analysis from a pro who can explain if it's something Mozilla should do better, or if this is the website which does something it shouldn't. I know it's low priority, but at least it's a test case for someone interested.

My turn: you give me advices but have you tried to reproduce it yourself ?

Comment 6

[Valentin Gosu [:valentin] (he/him) {{ AFK in July }}]

This might have something to do with the tailing code - since twitter frames usually trip tracking protection classifiers.

Comment 8

[Nojevah]

Pls, read my original report. I've explained there's a difference between Firefox and Chromium browser. That's why I've reported this ! So you can eliminate Internet speed, DNS or whatever. Just do the test and you'll see by yourself.
I don't blame Firefox engine because I would not be surprised if it's a problem with Twitter. But this report is useful I think.

Comment 9

[Andrew Creskey [:acreskey]]

Sorry so slow into getting to this.

We just completed significant upgrades to HTTP prioritization in Fx 132, including bug 1865040 and bug 1797715.

So I'm looking at this with the latest nightly, Fx 132.

I'll attach videos from Firefox for Android, and it does appear that Chrome is a little bit faster in getting the twitter.com iframes loaded (but I'm not seeing a huge difference).

But Valentin is correct in comment 6 -- the twitter iframe is identified as a tracker, and is thus scheduled after other network requests.
This looks to be the particular url:
https://platform.twitter.com/widgets/widget_iframe.2f70fb173b9000da126c79afe2098f02.html?origin=https%3A%2F%2Fwww.lequipe.fr

We're looking at tuning how this system defers requests in bug 1915866.

Comment 10

[Andrew Creskey [:acreskey]]

Attachment: fenix_iframe.mp4

Firefox for Android loading twitter iframe

Comment 11

[Andrew Creskey [:acreskey]]

Attachment: chrome_iframe.mp4

Chrome on Android loading twitter iframe

Comment 12

[Nojevah]

Thanks for this detailed answer. I'll check the progress of all mentionned bugs from time to time and report if perfs are identical the day it lands on official release.

(I see comment #7 has been deleted, so my comment #8 makes no sense, just to be sure nobody would think I was adressing to Valentin)

Comment 13

[aamernja]

(In reply to Andrew Creskey [:acreskey] from comment #11)

Created attachment 9427791 [details]
chrome_iframe.mp4

Chrome on Android loading twitter iframe

can you please explain in detail please or any other resources. please share it with me

Comment 14

[Nojevah]

What do you want him to explain ? It's already nice of him to share these videos showing there's no difference anymore between Chrome and Firefox now (hoping it's the case on desktop too).
But maybe you're just a new spam account and I'm too naive.

Comment 16

[Valentin Gosu [:valentin] (he/him) {{ AFK in July }}]

I think these are just spam

Comment 17

[Nojevah]

Good news [:valentin] & [:acreskey] ! I can confirm in Firefox 132 that frames are charging very quickly ! Thank you for your efforts. I let you mark this as resolved when you feel it !

Comment 19

[Andrew Creskey [:acreskey]]

(In reply to Muhammad Ali from comment #18)

I'm almost facing the same concerns when I open my client's vintage bracelet store. Some products don't load properly even though the page is properly loaded. When I open with another browser the speed is slightly better. How can this be fix?

Hi Muhammad -- I had a look at the site you listed, https://thebackvault.com/collections/vintage-bracelets.
I don't see any twitter iframes being delayed nor images being classified as trackers so this looks like a separate issue.

Can you file a new bug for that one, under Core::Performance?

If you can provide a performance profile that captures the problem, that would be helpful:
Tool: https://profiler.firefox.com/
Docs: https://profiler.firefox.com/docs/#/

Comment 20

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

Redirect a needinfo that is pending on an inactive user to the triage owner.
:kershaw, since the bug has recent activity, could you have a look please?

For more information, please visit BugBot documentation.

Comment 21

[Kershaw Chang [:kershaw]]

(In reply to Nojevah from comment #17)

Good news [:valentin] & [:acreskey] ! I can confirm in Firefox 132 that frames are charging very quickly ! Thank you for your efforts. I let you mark this as resolved when you feel it !

This seems to be fixed.
I'll close this bug. Please reopen if this happens again.