From 3d3e7f6391d8e47f62ede815f6e53956d14cd1eb Mon Sep 17 00:00:00 2001 From: Andrei Litvin Date: Fri, 4 Oct 2024 08:30:40 -0400 Subject: [PATCH] one more decoupling update: sligtly more inefficient, however likely this is not important --- src/app/CommandHandlerImpl.cpp | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/src/app/CommandHandlerImpl.cpp b/src/app/CommandHandlerImpl.cpp index 7f736dae1c0aae..f5eb159838111d 100644 --- a/src/app/CommandHandlerImpl.cpp +++ b/src/app/CommandHandlerImpl.cpp @@ -1,6 +1,6 @@ /* * - * Copyright (c) 2020-2024 Project CHIP Authors + * Copyright (c) 2020 Project CHIP Authors * All rights reserved. * * Licensed under the Apache License, Version 2.0 (the "License"); @@ -479,11 +479,19 @@ Status CommandHandlerImpl::ProcessGroupCommandDataIB(CommandDataIB::Parser & aCo // once up front and discard all the paths at once. Ordering with respect // to ACL and command presence checks does not matter, because the behavior // is the same for all of them: ignore the path. +#if !CHIP_CONFIG_USE_DATA_MODEL_INTERFACE + + // Without data model interface, we can query individual commands. + // Data model interface queries commands by a full path so we need endpointID as well. + // + // Since this is a performance update and group commands are never timed, + // missing this should not be that noticeable. if (CommandNeedsTimedInvoke(clusterId, commandId)) { // Group commands are never timed. return Status::Success; } +#endif // No check for `CommandIsFabricScoped` unlike in `ProcessCommandDataIB()` since group commands // always have an accessing fabric, by definition.