Inconsistent coding style for defining local variables

[ipython3]

Take the function iox::runtime::IpcRuntimeInterface::create() as an example.

expected<IpcRuntimeInterface, IpcRuntimeInterfaceError> IpcRuntimeInterface::create(const RuntimeName_t& runtimeName, const DomainId domainId, const units::Duration roudiWaitingTimeout) noexcept
{
    if (runtimeName.empty())
    {
        IOX_LOG(Debug, "The runtime name must not be empty!");
        return err(IpcRuntimeInterfaceError::CANNOT_CREATE_APPLICATION_CHANNEL);
    }

    MgmtShmCharacteristics mgmtShmCharacteristics;

    auto roudiIpcInterface = IpcInterfaceUser(roudi::IPC_CHANNEL_ROUDI_NAME, domainId, ResourceType::ICEORYX_DEFINED);

    auto appIpcInterface = IpcInterfaceCreator::create(runtimeName, domainId, ResourceType::USER_DEFINED);
    if (appIpcInterface.has_error() || !appIpcInterface->isInitialized())
    {
        return err(IpcRuntimeInterfaceError::CANNOT_CREATE_APPLICATION_CHANNEL);
    }

    deadline_timer timer(roudiWaitingTimeout);

We can see local variable defined like this:

// since C++17, a prvalue is directly constructed into the storage of its final destination
auto roudiIpcInterface = IpcInterfaceUser(roudi::IPC_CHANNEL_ROUDI_NAME, domainId, ResourceType::ICEORYX_DEFINED);

// direct initialization
deadline_timer timer(roudiWaitingTimeout);

As far as I know, since C++17, auto roudiIpcInterface = IpcInterfaceUser(roudi::IPC_CHANNEL_ROUDI_NAME, domainId, ResourceType::ICEORYX_DEFINED); is the same as IpcInterfaceUser roudiIpcInterface(roudi::IPC_CHANNEL_ROUDI_NAME, domainId, ResourceType::ICEORYX_DEFINED); . (Is that correct? Please correct me if I'm wrong)

So why we mix the two styles here? What makes us choose to use the first style when initializing roudiIpcInterface and the second style when initializing the timer?

[elBoberido]

@ipython3 iceoryy is an old codebase. It's roots are from 2014/2015. This means that many people worked on the codebase and it moved from C++11 to C++17.

This lead to quite a few inconsistencies. It's worse when you compare iceoryx_posh with iceoryx_hoofs. The latter mostly follows the STL style.

With iceoryx2 we did a better job and use clang-tidy right from the beginning.

[ipython3]

@elBoberido Thanks for sharing some of the project's history! I'm just reading iceoryx's code for C++ educational purposes, and raising this issue simply to confirm my understanding of this part of C++ syntax. So there's no particular reason for choosing between these two styles, just historical reasons, right?

[elBoberido]

@ipython3 yeah, historical reasons. You probanly will find this inconsistencies in most codebases with a lot of contributors and a long history.

Personally, I prefer auto to stick since it also prevents the most vexing parse issue.

Can this issue be closed?