Bug 164993 – REGRESSION (Safari 10): onbeforeunload does not show up after the first time Leave is clicked

• Status: RESOLVED
• Resolution: INVALID
• Component: Page Loading

[Manuel Perez]

Created attachment 295272 [details]
a test page that demonstrates and explains the problem

After the first time the user is presented with the dialog triggered by onbeforeunload and the user chooses to leave, the dialog does not show up any more regardless of the way used to navigate out of the page.

We found this issue on Safari 10.0 (12602.1.50.0.10) and then we were able to reproduce it on Version 10.0.1 and the latest nightly build.

See the attachment for complete description and steps to follow to reproduce the issue.

[Alexey Proskuryakov]

Do you know if this is a regression? In other words, did this issue occur with Safari 9?

In the attached test case, I assume that one needs to press the Back button between steps 2 and 3.

[Manuel Perez]

> Do you know if this is a regression? In other words, did this issue occur with Safari 9?

We just tested it and the issue cannot be reproduced on Safari 9.

> In the attached test case, I assume that one needs to press the Back button between steps 2 and 3.

If you test it directly in https://bug-164993-attachments.webkit.org/attachment.cgi?id=295272 without downloading the attachment then yes.

[Brady Eidson]

This is the back/forward cache, as I'm sure Alexey suspected.

[Manuel Perez]

Created attachment 295400 [details]
page 1 out of 2 to better demonstrate the problem

[Manuel Perez]

Created attachment 295402 [details]
page 2 out of 2 to better demonstrate the problem

[Manuel Perez]

Not sure it's related to the back/forward cache.

I've just uploaded a couple of attachments to demonstrate the issue without requiring to hit the back button. This time you'll have to download the pages in order to check the issue out.

Thanks in advance.

[Radar WebKit Bug Importer]

<rdar://problem/29464654>

[Daniel Bates]

This is a Safari bug. That is, it is not a WebKit bug. Marking this bug Resolved Invalid.

[Manuel Perez]

I think this is not a safari specific issue as it can be reproduced using WebKit-SVN-r208586.dmg

[Daniel Bates]

(In reply to comment #8)
> This is a Safari bug. That is, it is not a WebKit bug. Marking this bug
> Resolved Invalid.

For completeness the Safari bug affects Safari version 9.1 (601.5.17) and latter.

[Daniel Bates]

(In reply to comment #9)
> I think this is not a safari specific issue as it can be reproduced using
> WebKit-SVN-r208586.dmg

Notice that the nightlies use system Safari. What version of Safari are you using?

[Daniel Bates]

(In reply to comment #11)
> (In reply to comment #9)
> > I think this is not a safari specific issue as it can be reproduced using
> > WebKit-SVN-r208586.dmg
> 
Are you saying that you were able to reproduce this issue with Safari version < 9.1 (601.5.17) and WebKit-SVN-r208586.dmg?

[Edward Rudd]

(In reply to comment #12)
> (In reply to comment #11)
> > (In reply to comment #9)
> > > I think this is not a safari specific issue as it can be reproduced using
> > > WebKit-SVN-r208586.dmg
> > 
> Are you saying that you were able to reproduce this issue with Safari
> version < 9.1 (601.5.17) and WebKit-SVN-r208586.dmg?

We will check though our QA systems with older Safari and report back.. If it is indeed a safari only issue.  Where should we report that?

[Daniel Bates]

(In reply to comment #13)
> Where should we report that?

No need to report it again. We are internally tracking this in the radar provided in comment 7.

[Manuel Perez]

> Notice that the nightlies use system Safari. What version of Safari are you using?

I tested and reproduced the issue with Safari "Version 10.0.1 (12602.2.14.0.7, r208586)" and also with "Version 10.0.1 (12602.2.14.0.7, r209592)”.

> Are you saying that you were able to reproduce this issue with Safari version < 9.1 (601.5.17) and WebKit-SVN-r208586.dmg?

A teammate tested it on Safari Version 9.0 (11601.1.56) with Webkit Nightly r209601 and he could NOT reproduce the issue there.