Bug 192270 – Let JSDOMGlobalObject in uniqueIDBDatabase be deleted if no database is using it

• Status: RESOLVED
• Resolution: DUPLICATE
• Component: New Bugs

[Sihui Liu]

JSDOMGlobalObject in uniqueIDBDatabase is used for serializing and deserializing JavaScript values, so we don't have to keep it and its VM forever if there are no objects using it.

[Sihui Liu]

Created attachment 356268 [details]
Patch

[EWS Watchlist]

Comment on attachment 356268 [details]
Patch

Attachment 356268 [details] did not pass mac-debug-ews (mac):
Output: https://webkit-queues.webkit.org/results/10224031

New failing tests:
imported/w3c/IndexedDB-private-browsing/idbdatabase_createObjectStore10-1000ends.html

[EWS Watchlist]

Created attachment 356284 [details]
Archive of layout-test-results from ews117 for mac-sierra

The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews117  Port: mac-sierra  Platform: Mac OS X 10.12.6

[Chris Dumez]

Comment on attachment 356268 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=356268&action=review

layout test failure looks likely related.

> Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:90
> +    , m_serializationContext(nullptr)

Not needed, the RefPtr<> default constructor does the right thing.

> Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:967
> +void UniqueIDBDatabase::sharedIDBSerializationContext(RefPtr<IDBSerializationContext>& context)

Why isn't this returning the context instead of using an out-parameter? This looks confusing.

> Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:974
> +    if (!s_context.get().get())

Doesn't "if (!s_context)" work? At the very least "if (!s_context.get())" should.

> Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:1032
> +        sharedIDBSerializationContext(m_serializationContext);

This looks super confusing.

> Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.h:80
> +    IDBSerializationContext();

Constructor should be private and this class should have a create() factory static function. This is the convention for RefCounted classes. Also, I'd swap the base classes order.

[Sihui Liu]

(In reply to Chris Dumez from comment #4)
> Comment on attachment 356268 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=356268&action=review
> 
> layout test failure looks likely related.
> 
Fixed.

> > Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:90
> > +    , m_serializationContext(nullptr)
> 
> Not needed, the RefPtr<> default constructor does the right thing.
> 
Removed.

> > Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:967
> > +void UniqueIDBDatabase::sharedIDBSerializationContext(RefPtr<IDBSerializationContext>& context)
> 
> Why isn't this returning the context instead of using an out-parameter? This
> looks confusing.
> 
Changed to use a return value.

> > Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:974
> > +    if (!s_context.get().get())
> 
> Doesn't "if (!s_context)" work? At the very least "if (!s_context.get())"
> should.
> 
s_context.get() works!

> > Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:1032
> > +        sharedIDBSerializationContext(m_serializationContext);
> 
> This looks super confusing.
> 
Ditto.

> > Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.h:80
> > +    IDBSerializationContext();
> 
> Constructor should be private and this class should have a create() factory
> static function. This is the convention for RefCounted classes. Also, I'd
> swap the base classes order.
Done.

[Sihui Liu]

Created attachment 356426 [details]
Patch

[Geoffrey Garen]

Comment on attachment 356426 [details]
Patch

r=me

[Sihui Liu]

Probably a more complete solution in https://bugs.webkit.org/show_bug.cgi?id=200526.

*** This bug has been marked as a duplicate of bug 200526 ***