Bug 42863 – REGRESSION (r60104): Zoom level is unexpectedly reset on page reload

• Component: Page Loading
• Status: RESOLVED FIXED

[Xiaomei Ji]

Version: WebKit r63854 (was built on 21 July 2010 and is a 27.9 MB download)

Steps to reproduce:
1. open any webpage, for example: www.apple.com/startpage
2. continue to zoom in the page to the largest. 
4. reload the page

The page appears to be reset to normal size.
But the zoom level is preserved as the largest zoom level.  (the "zoom in" is greyed, and press zoom in wont work. But press zoom out will work which actually decreases the font size from the one before reload, and the font size appears to be bigger comparing to the one before zoom out).


Tested in: OSX 10.6

[Alexey Proskuryakov]

In shipping Safari/WebKit 5.0, the page stays zoomed in.

[Alexey Proskuryakov]

<rdar://problem/8231917>

[Matt Lilek]

*** Bug 41573 has been marked as a duplicate of this bug. ***

[Matt Lilek]

This regressed in http://trac.webkit.org/changeset/60104

[Darin Adler]

I’m surprised that moving these functions caused this. I guess a new FrameView is created and does not inherit the zoom level from the older FrameView?

[Darin Adler]

Created attachment 67963 [details]
Patch

[Daniel Bates]

We should be able to test this change by using a render tree dump/pixel test and eventSender.zoomPageIn().

[Daniel Bates]

Created attachment 67977 [details]
Layout Test

I was not sure where to place this test case. I noticed other test cases that used eventSender.zoomInPage() in fast/css. So, I put the test case in this directory. I am open to suggestions.

[Daniel Bates]

(In reply to comment #8)
> Created an attachment (id=67977) [details]
> Layout Test
> 
> I was not sure where to place this test case. I noticed other test cases that used eventSender.zoomInPage() in fast/css. So, I put the test case in this directory. I am open to suggestions.

I mean't to write eventSender.zoomPageIn() instead of eventSender.zoomInPage().

[Darin Adler]

Comment on attachment 67977 [details]
Layout Test

I’m surprised this works, but it’s good news that it does.

[Darin Adler]

Committed r67762: <http://trac.webkit.org/changeset/67762>

[WebKit Review Bot]

http://trac.webkit.org/changeset/67762 might have broken Qt Linux Release minimal, Qt Linux ARMv5 Release, and Qt Linux ARMv7 Release

[Dimitri Glazkov (Google)]

(In reply to comment #12)
> http://trac.webkit.org/changeset/67762 might have broken Qt Linux Release minimal, Qt Linux ARMv5 Release, and Qt Linux ARMv7 Release

All Chromium builds are borked :(

[Eric Seidel (no email)]

This appears to have torched the tree. :(

[Mihai Parparita]

I've put up a patch at bug 46040 that should at least fix the Snow Leopard bot (and unblock the commit queue) if someone wouldn't mind r+-ing and landing it for me.

[Darin Adler]

Sorry about the incorrect logic and build breakage. Hope everything’s fixed now. Thanks to those who helped patch things up.