Issue mach lookup extension to "com.apple.lsd.open" for Mail, since this service will be removed from the WebContent sandbox.
<rdar://problem/57990991>
Created attachment 385836 [details] Patch
I am also looking into creating a test for this.
Created attachment 386128 [details] Patch
Comment on attachment 386128 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=386128&action=review > LayoutTests/fast/sandbox/ios/sandbox-mach-lookup-mail.html:1 > +<!DOCTYPE html><!-- webkit-test-runner [ applicationBundleIdentifier=com.apple.mobilemail ] --> Cool!
Created attachment 386772 [details] Patch
Comment on attachment 386772 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=386772&action=review R=me > Tools/WebKitTestRunner/TestController.cpp:627 > + // Exit if the application bundle identifier has already been set, since it can only be set once. Could this just be a RELEASE_ASSERT?
(In reply to Brent Fulgham from comment #7) > Comment on attachment 386772 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=386772&action=review > > R=me > > > Tools/WebKitTestRunner/TestController.cpp:627 > > + // Exit if the application bundle identifier has already been set, since it can only be set once. > > Could this just be a RELEASE_ASSERT? Having a RELEASE_ASSERT when this happens, would then show up as a test crash, I think. Exiting will let the next WebKitTestRunner continue where the one that exited left off. Thanks for reviewing!
Comment on attachment 386772 [details] Patch Clearing flags on attachment: 386772 Committed r254052: <https://trac.webkit.org/changeset/254052>
All reviewed patches have been landed. Closing bug.
Why is this being done this way? Client identity checks are necessary when the client is some existing code that’s not part of the system and can’t be updated. In this case, however, Mail can either be provided with explicit API for opting into this behavior, or it can probably just as easily restructure its code such that the interaction with lsd happens in the UI process.
(In reply to mitz from comment #11) > Why is this being done this way? Client identity checks are necessary when > the client is some existing code that’s not part of the system and can’t be > updated. In this case, however, Mail can either be provided with explicit > API for opting into this behavior, or it can probably just as easily > restructure its code such that the interaction with lsd happens in the UI > process. That is an excellent idea, assuming we can get time on Mail's schedule to adopt such an API. We should proceed with this first step, and open a task to create relevant API and track adoption, at which point we could remove this internal client check.