Resource Load Statistics (experimental): Delete non-cookie website data after 7 days of no user interaction

• Component: WebKit Misc.
• Status: RESOLVED FIXED

[John Wilander]

Trackers are continuing to move cross-site tracking IDs into first-party storage. We should age out script-writable non-cookie website data in alignment with the 7-day cap on client-side cookies.

[Radar WebKit Bug Importer]

<rdar://problem/57578989>

[John Wilander]

Created attachment 384681 [details]
Patch

[John Wilander]

The Win test failures are all over the place and seem unrelated.

[Alex Christensen]

Comment on attachment 384681 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=384681&action=review

> Source/WebKit/NetworkProcess/Classifier/WebResourceLoadStatisticsStore.cpp:488
> +    postTask([this, mode, completionHandler = WTFMove(completionHandler)]() mutable {

Not for this patch, but I don't like it how in this one file it's ok to use this without protecting it or checking a weak pointer, but everywhere else we need to be careful.  Let's make it so postTask doesn't implicitly protect this and require all callers to protect this.

[Chris Dumez]

(In reply to Alex Christensen from comment #4)
> Comment on attachment 384681 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=384681&action=review
> 
> > Source/WebKit/NetworkProcess/Classifier/WebResourceLoadStatisticsStore.cpp:488
> > +    postTask([this, mode, completionHandler = WTFMove(completionHandler)]() mutable {
> 
> Not for this patch, but I don't like it how in this one file it's ok to use
> this without protecting it or checking a weak pointer, but everywhere else
> we need to be careful.  Let's make it so postTask doesn't implicitly protect
> this and require all callers to protect this.

Previously it was like that and people kept forgetting to protect this and we had tons of similar crashes. As a result, I made postTask() protect this. Worse case scenario, the caller protects again, which is a bit inefficient but at least does not cause crashes.

[John Wilander]

Thanks for the review!

[WebKit Commit Bot]

Comment on attachment 384681 [details]
Patch

Clearing flags on attachment: 384681

Committed r253082: <https://trac.webkit.org/changeset/253082>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.