Bug 227220 – [WK2] Don't process-swap on navigations within the same non-HTTP(s) protocol

• Component: WebKit2
• Status: RESOLVED FIXED

[Chris Dumez]

Don't process-swap on navigations within the same non-HTTP(s) protocol. Trying to extract registrable domains from non HTTP(s) URLs does not make much sense and leads to unexpected process swaps.

[Chris Dumez]

Created attachment 431870 [details]
Patch

[Chris Dumez]

<rdar://79106461>

[Chris Dumez]

Created attachment 431877 [details]
Patch

[EWS]

Committed r279079 (238998@main): <https://commits.webkit.org/238998@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 431877 [details].

[Aakash Jain]

(In reply to EWS from comment #4)
> Committed r279079 (238998@main): <https://commits.webkit.org/238998@main>
Seems like it broke api-gtk test /WebKit2Gtk/TestWebExtensions:/webkit/WebKitWebExtension/page-id

[Chris Dumez]

(In reply to Aakash Jain from comment #5)
> (In reply to EWS from comment #4)
> > Committed r279079 (238998@main): <https://commits.webkit.org/238998@main>
> Seems like it broke api-gtk test
> /WebKit2Gtk/TestWebExtensions:/webkit/WebKitWebExtension/page-id

I'll see if I can figure it out but did we ever change the rule that WK2 changes were allowed to break non-Cocoa ports to not hold back development?

[Chris Dumez]

The GTK test is observing process swaps by checking if the pageID changed or not. Should be easy to update the test. Looking into it now.

[Chris Dumez]

Reopening to attach new patch.

[Chris Dumez]

Created attachment 431917 [details]
Follow-up to fix GTK API test

[Aakash Jain]

> I'll see if I can figure it out but did we ever change the rule that WK2 changes were allowed to break non-Cocoa ports to not hold back development?
Thanks. I am not sure of the exact rule. However, this test breakage is slowing does https://ews-build.webkit.org/#/builders/API-Tests-GTK-EWS queue. This would result in progressively increasing delay in getting ews results on that queue, which might result in reduced coverage and thereby causing more regressions on that queue. From maintaining the infrastructure point of view, it's important to keep the test-suites in good state.

[Chris Dumez]

(In reply to Aakash Jain from comment #10)
> > I'll see if I can figure it out but did we ever change the rule that WK2 changes were allowed to break non-Cocoa ports to not hold back development?
> Thanks. I am not sure of the exact rule. However, this test breakage is
> slowing does https://ews-build.webkit.org/#/builders/API-Tests-GTK-EWS
> queue. This would result in progressively increasing delay in getting ews
> results on that queue, which might result in reduced coverage and thereby
> causing more regressions on that queue. From maintaining the infrastructure
> point of view, it's important to keep the test-suites in good state.

Understood. I think breaking other ports is bad practice either way, I was just surprised to get ping'd by Apple about it.

I have a speculative fix ready but am waiting for EWS to process and confirm it works since I cannot validate locally.

[EWS]

Committed r279095 (239012@main): <https://commits.webkit.org/239012@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 431917 [details].