Bug 238583 – Drop bad adoptNS() in NetworkProcess.LaunchOnlyWhenNecessary

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Chris Dumez]

Drop bad adoptNS() in NetworkProcess.LaunchOnlyWhenNecessary. It is incorrect as the WKWebViewConfiguration.websiteDataStore getter is not ref'ing the store for us.

[Chris Dumez]

Created attachment 456190 [details]
Patch

[Alex Christensen]

If only there were some kind of language feature we could use that would do all this for us...

[Darin Adler]

A sort of automatic release counting system?

[Chris Dumez]

(In reply to Darin Adler from comment #3)
> A sort of automatic release counting system?

Do we have ARC enabled in the API tests?

[Darin Adler]

I don’t think we do, but I think we could.

Thinking about it, it seems to me that calls to adoptNS should fail to compile under ARC; not important to be able to write it. Should be easy enough to make that change to RetainPtr.h if it’s not already done. Then the question wouldn’t need to be asked. If this was under ARC then it would fail to compile.

A while back, once WebKit had a new enough minimum supported version of iOS and macOS, I tried to start converting everything to ARC and it was hard to get it all done. My patches introduced storage leaks because presumably I created a reference cycle somewhere by accident. But probably it wasn’t as hard as I made it. It seems safer for the future if we could get over that hump and convert all our code.

Should be easy to switch over all our tests, though, and less risky than doing it in the framework code. So maybe we should do that really soon.

[EWS]

Committed r292144 (249051@main): <https://commits.webkit.org/249051@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 456190 [details].

[Radar WebKit Bug Importer]

<rdar://problem/91091275>