feedc0de/sdbus-cpp
1
0
Fork 0
forked from Kistler-Group/sdbus-cpp
Code Pull Requests Activity

Move C++17 uncaught_exceptions to public API

Browse Source
This commit is contained in:
sangelovic
2019-11-10 17:31:58 +01:00
parent 3a76e9c120
commit d50a15b2a2
6 changed files with 48 additions and 278 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/ConvenienceApiClasses.cpp
-209
View File
@@ -1,209 +0,0 @@
/**
* (C) 2016 - 2017 KISTLER INSTRUMENTE AG, Winterthur, Switzerland
* (C) 2016 - 2019 Stanislav Angelovic <angelovic.s@gmail.com>
*
* @file ConvenienceApiClasses.cpp
*
* Created on: Jan 19, 2017
* Project: sdbus-c++
* Description: High-level D-Bus IPC C++ library based on sd-bus
*
* This file is part of sdbus-c++.
*
* sdbus-c++ is free software; you can redistribute it and/or modify it
* under the terms of the GNU Lesser General Public License as published by
* the Free Software Foundation, either version 2.1 of the License, or
* (at your option) any later version.
*
* sdbus-c++ is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with sdbus-c++. If not, see <http://www.gnu.org/licenses/>.
*/
#include "sdbus-c++/ConvenienceApiClasses.h"
#include "sdbus-c++/IObject.h"
#include "sdbus-c++/IProxy.h"
#include <string>
#include <exception>
namespace sdbus {
MethodRegistrator::MethodRegistrator(IObject& object, const std::string& methodName)
: object_(object)
, methodName_(methodName)
, exceptions_(std::uncaught_exceptions()) // Needs C++17
{
}
MethodRegistrator::~MethodRegistrator() noexcept(false) // since C++11, destructors must
{ // explicitly be allowed to throw
// Don't register the method if MethodRegistrator threw an exception in one of its methods
if (std::uncaught_exceptions() != exceptions_)
return;
SDBUS_THROW_ERROR_IF(interfaceName_.empty(), "DBus interface not specified when registering a DBus method", EINVAL);
SDBUS_THROW_ERROR_IF(!methodCallback_, "Method handler not specified when registering a DBus method", EINVAL);
// registerMethod() can throw. But as the MethodRegistrator shall always be used as an unnamed,
// temporary object, i.e. not as a stack-allocated object, the double-exception situation
// shall never happen. I.e. it should not happen that this destructor is directly called
// in the stack-unwinding process of another flying exception (which would lead to immediate
// termination). It can be called indirectly in the destructor of another object, but that's
// fine and safe provided that the caller catches exceptions thrown from here.
// Therefore, we can allow registerMethod() to throw even if we are in the destructor.
// Bottomline is, to be on the safe side, the caller must take care of catching and reacting
// to the exception thrown from here if the caller is a destructor itself.
object_.registerMethod(interfaceName_, methodName_, inputSignature_, outputSignature_, std::move(methodCallback_), flags_);
}
SignalRegistrator::SignalRegistrator(IObject& object, const std::string& signalName)
: object_(object)
, signalName_(signalName)
, exceptions_(std::uncaught_exceptions()) // Needs C++17
{
}
SignalRegistrator::~SignalRegistrator() noexcept(false) // since C++11, destructors must
{ // explicitly be allowed to throw
// Don't register the signal if SignalRegistrator threw an exception in one of its methods
if (std::uncaught_exceptions() != exceptions_)
return;
SDBUS_THROW_ERROR_IF(interfaceName_.empty(), "DBus interface not specified when registering a DBus signal", EINVAL);
// registerSignal() can throw. But as the SignalRegistrator shall always be used as an unnamed,
// temporary object, i.e. not as a stack-allocated object, the double-exception situation
// shall never happen. I.e. it should not happen that this destructor is directly called
// in the stack-unwinding process of another flying exception (which would lead to immediate
// termination). It can be called indirectly in the destructor of another object, but that's
// fine and safe provided that the caller catches exceptions thrown from here.
// Therefore, we can allow registerSignal() to throw even if we are in the destructor.
// Bottomline is, to be on the safe side, the caller must take care of catching and reacting
// to the exception thrown from here if the caller is a destructor itself.
object_.registerSignal(interfaceName_, signalName_, signalSignature_, flags_);
}
PropertyRegistrator::PropertyRegistrator(IObject& object, const std::string& propertyName)
: object_(object)
, propertyName_(propertyName)
, exceptions_(std::uncaught_exceptions())
{
}
PropertyRegistrator::~PropertyRegistrator() noexcept(false) // since C++11, destructors must
{ // explicitly be allowed to throw
// Don't register the property if PropertyRegistrator threw an exception in one of its methods
if (std::uncaught_exceptions() != exceptions_)
return;
SDBUS_THROW_ERROR_IF(interfaceName_.empty(), "DBus interface not specified when registering a DBus property", EINVAL);
// registerProperty() can throw. But as the PropertyRegistrator shall always be used as an unnamed,
// temporary object, i.e. not as a stack-allocated object, the double-exception situation
// shall never happen. I.e. it should not happen that this destructor is directly called
// in the stack-unwinding process of another flying exception (which would lead to immediate
// termination). It can be called indirectly in the destructor of another object, but that's
// fine and safe provided that the caller catches exceptions thrown from here.
// Therefore, we can allow registerProperty() to throw even if we are in the destructor.
// Bottomline is, to be on the safe side, the caller must take care of catching and reacting
// to the exception thrown from here if the caller is a destructor itself.
object_.registerProperty( std::move(interfaceName_)
, std::move(propertyName_)
, std::move(propertySignature_)
, std::move(getter_)
, std::move(setter_)
, flags_ );
}
InterfaceFlagsSetter::InterfaceFlagsSetter(IObject& object, const std::string& interfaceName)
: object_(object)
, interfaceName_(interfaceName)
, exceptions_(std::uncaught_exceptions())
{
}
InterfaceFlagsSetter::~InterfaceFlagsSetter() noexcept(false) // since C++11, destructors must
{ // explicitly be allowed to throw
// Don't set any flags if InterfaceFlagsSetter threw an exception in one of its methods
if (std::uncaught_exceptions() != exceptions_)
return;
SDBUS_THROW_ERROR_IF(interfaceName_.empty(), "DBus interface not specified when setting its flags", EINVAL);
// setInterfaceFlags() can throw. But as the InterfaceFlagsSetter shall always be used as an unnamed,
// temporary object, i.e. not as a stack-allocated object, the double-exception situation
// shall never happen. I.e. it should not happen that this destructor is directly called
// in the stack-unwinding process of another flying exception (which would lead to immediate
// termination). It can be called indirectly in the destructor of another object, but that's
// fine and safe provided that the caller catches exceptions thrown from here.
// Therefore, we can allow setInterfaceFlags() to throw even if we are in the destructor.
// Bottomline is, to be on the safe side, the caller must take care of catching and reacting
// to the exception thrown from here if the caller is a destructor itself.
object_.setInterfaceFlags( std::move(interfaceName_)
, std::move(flags_) );
}
SignalEmitter::SignalEmitter(IObject& object, const std::string& signalName)
: object_(object)
, signalName_(signalName)
, exceptions_(std::uncaught_exceptions()) // Needs C++17
{
}
SignalEmitter::~SignalEmitter() noexcept(false) // since C++11, destructors must
{ // explicitly be allowed to throw
// Don't emit the signal if SignalEmitter threw an exception in one of its methods
if (std::uncaught_exceptions() != exceptions_)
return;
SDBUS_THROW_ERROR_IF(!signal_.isValid(), "DBus interface not specified when emitting a DBus signal", EINVAL);
// emitSignal() can throw. But as the SignalEmitter shall always be used as an unnamed,
// temporary object, i.e. not as a stack-allocated object, the double-exception situation
// shall never happen. I.e. it should not happen that this destructor is directly called
// in the stack-unwinding process of another flying exception (which would lead to immediate
// termination). It can be called indirectly in the destructor of another object, but that's
// fine and safe provided that the caller catches exceptions thrown from here.
// Therefore, we can allow emitSignal() to throw even if we are in the destructor.
// Bottomline is, to be on the safe side, the caller must take care of catching and reacting
// to the exception thrown from here if the caller is a destructor itself.
object_.emitSignal(signal_);
}
MethodInvoker::MethodInvoker(IProxy& proxy, const std::string& methodName)
: proxy_(proxy)
, methodName_(methodName)
, exceptions_(std::uncaught_exceptions()) // Needs C++17
{
}
MethodInvoker::~MethodInvoker() noexcept(false) // since C++11, destructors must
{ // explicitly be allowed to throw
// Don't call the method if it has been called already or if MethodInvoker
// threw an exception in one of its methods
if (methodCalled_ || std::uncaught_exceptions() != exceptions_)
return;
SDBUS_THROW_ERROR_IF(!method_.isValid(), "DBus interface not specified when calling a DBus method", EINVAL);
// callMethod() can throw. But as the MethodInvoker shall always be used as an unnamed,
// temporary object, i.e. not as a stack-allocated object, the double-exception situation
// shall never happen. I.e. it should not happen that this destructor is directly called
// in the stack-unwinding process of another flying exception (which would lead to immediate
// termination). It can be called indirectly in the destructor of another object, but that's
// fine and safe provided that the caller catches exceptions thrown from here.
// Therefore, we can allow callMethod() to throw even if we are in the destructor.
// Bottomline is, to be on the safe side, the caller must take care of catching and reacting
// to the exception thrown from here if the caller is a destructor itself.
proxy_.callMethod(method_, timeout_);
}
}
src/Object.cpp
+2
View File
@@ -134,6 +134,8 @@ sdbus::Signal Object::createSignal(const std::string& interfaceName, const std::
void Object::emitSignal(const sdbus::Signal& message)
{
SDBUS_THROW_ERROR_IF(!message.isValid(), "Invalid signal message provided", EINVAL);
message.send();
}
src/Proxy.cpp
+4
View File
@@ -71,11 +71,15 @@ AsyncMethodCall Proxy::createAsyncMethodCall(const std::string& interfaceName, c
MethodReply Proxy::callMethod(const MethodCall& message, uint64_t timeout)
{
SDBUS_THROW_ERROR_IF(!message.isValid(), "Invalid method call message provided", EINVAL);
return message.send(timeout);
}
void Proxy::callMethod(const AsyncMethodCall& message, async_reply_handler asyncReplyCallback, uint64_t timeout)
{
SDBUS_THROW_ERROR_IF(!message.isValid(), "Invalid async method call message provided", EINVAL);
auto callback = (void*)&Proxy::sdbus_async_reply_handler;
auto callData = std::make_unique<AsyncCalls::CallData>(AsyncCalls::CallData{*this, std::move(asyncReplyCallback), {}});