Bug 182598 – REGRESSION (STP 49): After opening a link in a new tab, the page center is off at first

• Component: Layout and Rendering
• Status: RESOLVED MOVED

[Tobi Reif]

Created attachment 333363 [details]
screen recording

After opening a link in a new tab, the page center is off at first (or perhaps the new (hidden) tab is very small at first?)

In Safari, open https://tobireif.com/demos/ . Open the Swirls demo in a new tab (eg using cmd-click). The page opens in (in a new tab in the foreground).
The swirls originate from the center.

In Safari Tech Preview, open https://tobireif.com/demos/ . Open the Swirls demo in a new tab (eg using cmd-click). The page opens in (in a new tab in the background - that's OK). Click on that tab.
The swirls at first don't originate from the center (as intended), instead they originate from a place near top left. Only after that they originate from the center (as intended).

(Also, sometimes clicking on a tab to switch to that tab doesn't work.)

[Tobi Reif]

https://bugs.webkit.org/attachment.cgi?id=333363&action=edit
"Attachment is not viewable in your browser because its MIME type (video/quicktime) is not one that your browser is able to display."

Perhaps .mov attachments should be supported?

[Simon Fraser (smfr)]

This is new in STP 49?

[Simon Fraser (smfr)]

Tobi, what math is the demo using to place the swirls? Is it just looking at innerWidth/innerHeight?

[Radar WebKit Bug Importer]

<rdar://problem/37358863>

[Simon Fraser (smfr)]

This is a regression in Safari code, not WebKit code. We'll track it internally.

[Tobi Reif]

Hi Simon!

Thanks for investigating.

> what math is the demo using to place the swirls? Is it just looking at
> innerWidth/innerHeight?

This code creates & adds the WegGL canvas element:

var renderer = PIXI.autoDetectRenderer(
  window.innerWidth, window.innerHeight
);
document.body.insertBefore(
  renderer.view, document.body.firstChild
);

This code positions each swirl:

[snipped].sprite.position.x = renderer.width / 2;
[snipped].sprite.position.y  = renderer.height / 2

So basically the answer is yes :)

[Simon Fraser (smfr)]

Cool, thanks.

[Tobi Reif]

As far as I can this this is fixed now.

[Tobi Reif]

Thanks!