IDBLevelDBCoding::m_metaDataType has been changed to unsigned so ASSERT(m_metaDataType >= 0) is not useful: it does create build problems at least with EFL. Attaching patch.
Created attachment 181896 [details] Patch
How about this? > ObjectStoreMetaDataKey::ObjectStoreMetaDataKey() > : m_objectStoreId(-1) > , m_metaDataType(-1) > { > }
Comment on attachment 181896 [details] Patch OK
(In reply to comment #2) > How about this? > > > ObjectStoreMetaDataKey::ObjectStoreMetaDataKey() > > : m_objectStoreId(-1) > > , m_metaDataType(-1) > > { > > } Good that you mentioned, I was planning to but forgot. It's ugly but should work, assuming that presents a metadatatype that does not exist: see "enum MetaDataType" in the header for the 'good' values of metadatatype. We could add a unsetObjectStoreMetaDataType but that shouldn't be part of this patch I guess.
Comment on attachment 181896 [details] Patch Please fix the issue Christophe pointed out in a follow-up patch.
Comment on attachment 181896 [details] Patch Clearing flags on attachment: 181896 Committed r139193: <http://trac.webkit.org/changeset/139193>
All reviewed patches have been landed. Closing bug.