Eliminate unnecessary hash lookups with NetworkResourceData There are a few places where we do multiple hash lookups on NetworkResourceData that we don't need: willSendRequest: (combine these into 1 call) m_resourcesData->resourceCreated(requestId, m_pageAgent->loaderId(&loader)); m_resourcesData->setResourceType(requestId, type); didReceiveResponse: (combine these into 1 call) m_resourcesData->responseReceived(requestId, m_pageAgent->frameId(loader.frame()), response); m_resourcesData->setResourceType(requestId, type); didLoadResourceFromMemoryCache: (combine these into 1 call) m_resourcesData->resourceCreated(requestId, loaderId); m_resourcesData->addCachedResource(requestId, &resource); Combining will eliminate a hash lookup in each case.
Created attachment 326187 [details] [PATCH] Proposed Fix
Comment on attachment 326187 [details] [PATCH] Proposed Fix View in context: https://bugs.webkit.org/attachment.cgi?id=326187&action=review > Source/WebCore/inspector/NetworkResourcesData.cpp:131 > + auto* resourceData = new ResourceData(requestId, loaderId); > + resourceData->setType(type); > + m_requestIdToResourceDataMap.set(requestId, resourceData); An additional cleanup that you might want to do is change m_requestIdToResourceDataMap to store a std::unique_ptr rather than a raw ptr for its value. > Source/WebCore/inspector/NetworkResourcesData.cpp:134 > +void NetworkResourcesData::resourceCreated(const String& requestId, const String& loaderId, CachedResource* cachedResource) Can this take the CachedResource by reference? > Source/WebCore/inspector/agents/InspectorNetworkAgent.cpp:351 > + String loaderId = m_pageAgent->loaderId(&loader); Can use auto. > Source/WebCore/inspector/agents/InspectorNetworkAgent.cpp:444 > + String frameId = m_pageAgent->frameId(loader.frame()); > + String loaderId = m_pageAgent->loaderId(&loader); Can use auto. > Source/WebCore/inspector/agents/InspectorNetworkAgent.cpp:531 > + String loaderId = m_pageAgent->loaderId(&loader); > + String frameId = m_pageAgent->frameId(loader.frame()); Can use auto.
Comment on attachment 326187 [details] [PATCH] Proposed Fix View in context: https://bugs.webkit.org/attachment.cgi?id=326187&action=review >> Source/WebCore/inspector/NetworkResourcesData.cpp:131 >> + m_requestIdToResourceDataMap.set(requestId, resourceData); > > An additional cleanup that you might want to do is change m_requestIdToResourceDataMap to store a std::unique_ptr rather than a raw ptr for its value. This is an excellent idea, I should have addressed this in this change anyways instead of allowing `new` / `delete` in our code. I'll put up a new patch. >> Source/WebCore/inspector/NetworkResourcesData.cpp:134 >> +void NetworkResourcesData::resourceCreated(const String& requestId, const String& loaderId, CachedResource* cachedResource) > > Can this take the CachedResource by reference? Yep.
(In reply to Joseph Pecoraro from comment #0) > Eliminate unnecessary hash lookups with NetworkResourceData > > There are a few places where we do multiple hash lookups on > NetworkResourceData that we don't need: Another multiple-hash-lookup we can remove is: ensureNoDataForRequestId ResourceData* resourceData = resourceDataForRequestId(requestId); ... m_requestIdToResourceDataMap.remove(requestId);
Created attachment 326233 [details] [PATCH] Proposed Fix
Attachment 326233 [details] did not pass style-queue: ERROR: Source/WebCore/inspector/NetworkResourcesData.cpp:273: More than one command on the same line [whitespace/newline] [4] Total errors found: 1 in 4 files If any of these errors are false positives, please file a bug against check-webkit-style.
Comment on attachment 326233 [details] [PATCH] Proposed Fix View in context: https://bugs.webkit.org/attachment.cgi?id=326233&action=review r=me > Source/WebCore/inspector/NetworkResourcesData.cpp:266 > + return entry.value->loaderId() != loaderId; v. cool
Comment on attachment 326233 [details] [PATCH] Proposed Fix Clearing flags on attachment: 326233 Committed r224596: <https://trac.webkit.org/changeset/224596>
All reviewed patches have been landed. Closing bug.
<rdar://problem/35562212>