There are currently a few compile and link errors when building WinCairo with IndexedDB enabled.
Created attachment 276664 [details] Patch
Created attachment 276667 [details] Patch
Attachment 276667 [details] did not pass style-queue: ERROR: Source/WebCore/PlatformWin.cmake:189: There should be exactly one empty line instead of 0 between "Modules/indexeddb" and "Modules/indexeddb/client". [list/emptyline] [5] ERROR: Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:1188: Place brace on its own line for function definitions. [whitespace/braces] [4] Total errors found: 2 in 12 files If any of these errors are false positives, please file a bug against check-webkit-style.
Comment on attachment 276667 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=276667&action=review cool! > Source/WebCore/Modules/indexeddb/server/MemoryBackingStoreTransaction.cpp:44 > - return std::make_unique<MemoryBackingStoreTransaction>(backingStore, info); > + return std::unique_ptr<MemoryBackingStoreTransaction>(new MemoryBackingStoreTransaction(backingStore, info)); This shouldn't be needed. > Source/WebCore/Modules/indexeddb/server/MemoryIDBBackingStore.cpp:48 > - return std::make_unique<MemoryIDBBackingStore>(identifier); > + return std::unique_ptr<MemoryIDBBackingStore>(new MemoryIDBBackingStore(identifier)); This shouldn't be needed. > Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:1188 > - auto callback = [this, self, refTransaction](const IDBError& error) { > + ErrorCallback callback = [this, self, refTransaction](const IDBError& error) { This shouldn't be needed. > Source/WebKit/win/storage/WebDatabaseProvider.cpp:45 > + sprintf_s(databaseDirectory, MAX_PATH, "%s\\%s", appDataDirectory, executableName); On Mac we use .../WebKit/Databases/___IndexedDB. We should probably do something similar here.
Comment on attachment 276667 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=276667&action=review > Source/WebKit/win/WebView.cpp:5062 > + RuntimeEnabledFeatures::sharedFeatures().setWebkitIndexedDBEnabled(true); This should probably be false by default, then set to true for DumpRenderTree and MiniBrowser.
(In reply to comment #4) > Comment on attachment 276667 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=276667&action=review > > cool! > Thanks for reviewing :) > > Source/WebCore/Modules/indexeddb/server/MemoryBackingStoreTransaction.cpp:44 > > - return std::make_unique<MemoryBackingStoreTransaction>(backingStore, info); > > + return std::unique_ptr<MemoryBackingStoreTransaction>(new MemoryBackingStoreTransaction(backingStore, info)); > > This shouldn't be needed. > I get a compile error without this: error C2248: 'WebCore::IDBServer::MemoryBackingStoreTransaction::MemoryBackingStoreTransaction': cannot access private member declared in class 'WebCore::IDBServer::MemoryBackingStoreTransaction Maybe we should fix this in another way? > > > Source/WebCore/Modules/indexeddb/server/UniqueIDBDatabase.cpp:1188 > > - auto callback = [this, self, refTransaction](const IDBError& error) { > > + ErrorCallback callback = [this, self, refTransaction](const IDBError& error) { > > This shouldn't be needed. > I get an error without this: error C2668: 'WebCore::IDBServer::UniqueIDBDatabase::storeCallback': ambiguous call to overloaded function > > Source/WebKit/win/storage/WebDatabaseProvider.cpp:45 > > + sprintf_s(databaseDirectory, MAX_PATH, "%s\\%s", appDataDirectory, executableName); > > On Mac we use .../WebKit/Databases/___IndexedDB. We should probably do > something similar here. I will update the path :)
Created attachment 276755 [details] Patch
Attachment 276755 [details] did not pass style-queue: ERROR: Source/WebCore/PlatformWin.cmake:189: There should be exactly one empty line instead of 0 between "Modules/indexeddb" and "Modules/indexeddb/client". [list/emptyline] [5] Total errors found: 1 in 12 files If any of these errors are false positives, please file a bug against check-webkit-style.
(In reply to comment #4) > Comment on attachment 276667 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=276667&action=review > > cool! > > > Source/WebCore/Modules/indexeddb/server/MemoryBackingStoreTransaction.cpp:44 > > - return std::make_unique<MemoryBackingStoreTransaction>(backingStore, info); > > + return std::unique_ptr<MemoryBackingStoreTransaction>(new MemoryBackingStoreTransaction(backingStore, info)); > > This shouldn't be needed. > I am afraid I didn't find a better way to fix these compile errors. The main problem seems to be that MSVC does not accept that std::make_unique is declared as a friend of a class.
Comment on attachment 276755 [details] Patch Just make the constructor public.
Created attachment 276760 [details] Patch
(In reply to comment #10) > Comment on attachment 276755 [details] > Patch > > Just make the constructor public. Thanks :) Updated patch.
Attachment 276760 [details] did not pass style-queue: ERROR: Source/WebCore/PlatformWin.cmake:189: There should be exactly one empty line instead of 0 between "Modules/indexeddb" and "Modules/indexeddb/client". [list/emptyline] [5] Total errors found: 1 in 12 files If any of these errors are false positives, please file a bug against check-webkit-style.
Comment on attachment 276760 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=276760&action=review Looks good. I think this needs another iteration, though. > Source/WebCore/PlatformWin.cmake:189 > Modules/indexeddb > + Modules/indexeddb/client Style bot wants a space here. > Source/WebCore/bindings/js/JSDOMWindowBase.h:23 > +#include "DOMWindow.h" Is this really necessary? What error does this prevent? Can we not forward declare things? Including more headers in a header like this makes the build slower. > Source/WebCore/bindings/js/JSDOMWrapper.h:25 > +#include "DOMWrapperWorld.h" Ditto. > Source/WebKit/win/storage/WebDatabaseProvider.cpp:41 > + return String(""); This isn't used in WebKit. I think this is what emptyString() is for. > Source/WebKit/win/storage/WebDatabaseProvider.cpp:50 > + return String(""); ditto.
Created attachment 276924 [details] Patch
(In reply to comment #14) > Comment on attachment 276760 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=276760&action=review > Thanks for reviewing! > > > Source/WebCore/bindings/js/JSDOMWindowBase.h:23 > > +#include "DOMWindow.h" > > Is this really necessary? What error does this prevent? Can we not forward > declare things? Including more headers in a header like this makes the > build slower. > I get the error: wtf/PassRefPtr.h(42): error C2027: use of undefined type 'WebCore::DOMWindow' PassRefPtr<DOMWindow> is used in JSDOMWindowBase.h, but this fails to compile since the class DOMWindow is only forward declared here.
Comment on attachment 276924 [details] Patch Clearing flags on attachment: 276924 Committed r200036: <http://trac.webkit.org/changeset/200036>
All reviewed patches have been landed. Closing bug.