Open Bug 1552713 Opened 7 months ago Updated 9 days ago

Site zoom is off with regards to the device width (regression)


(Core :: Layout, defect, P3)

68 Branch



Webcompat Priority P2
Tracking Status
firefox67 --- unaffected
firefox68 --- fix-optional


(Reporter: karlcow, Unassigned)


(Regression, )


(Keywords: regression)


(1 file)

  1. With Firefox Nightly on Android (68.0a1 2019-05-16)
  2. Go to

site displayed correctly like in Chrome latest version or Firefox 67 on Android

The site is zoomed in strange ways.

I will check what happens there later.

Flags: needinfo?(hikezoe)
Regressed by: 1500314

I spent some time to debug what Chrome actually does.

Chrome does pretty similar thing what we do and actually in the case of the site in question, result_width is the display size width. But after that a CSS style sheet which has @viewport { min-width: 980px } is loaded (I think it's a UA style sheet for Android), by this min-width extend-zoom happens, then the site is rendered properly.

We don't yet implement @viewport rule so we can't take the same approach now.

Flags: needinfo?(hikezoe)
See Also: → @viewport

I suspect that the issue might be happening because of the <meta name="viewport" content="user-scalable=no">. Tried removing the content attribute (right click on "content" in that meta tag -> Attributes -> Remove Attribute content ) and the site zoom is working the same as in Chrome. Tried this in remote debugging mode.

Attached a sample html that I was able to reproduce the issue with.

Yes, that's right. Before this change we had used 980px for layout if there is neither width and height in viewport meta tags. But after the change we use device display size for layout in such cases.

See Also: → 1556275
Webcompat Priority: ? → P2
See Also: → 1564104

invissionapp cases ( and look a different issue to me. I will check what's going on there in Chrome.

Flags: needinfo?(hikezoe.birchill)

Moving invisionapp cases to bug 1498729 since the site replaces the content of the same meta element.

Flags: needinfo?(hikezoe.birchill) is another case of what I commented in comment 2, thanks to Karl, he noticed if there is width=device-width, the behavior looks different, and it makes me notice that there is an @viewport rule in Chrome code.

You need to log in before you can comment on or make changes to this bug.