On 10/30/24 00:56, Frederick Virchanza Gotham via Std-Proposals wrote:
> On Tue, Oct 29, 2024 at 8:48 PM Andrey Semashev wrote:
>>
>>
>>> You could replace it with:
>>>
>>> string GetCodecName(Manager *const pmanager)
>>> {
>>> return pmanager->GetDevice()->GetChannel(0)->GetLowPassFilter()->GetCodecName();
>>> }
>>>
>>> That's the kind of lines I was thinking along.
>>
>> The two pieces of code are not equivalent, even assuming the proposal is
>> accepted. Specifically, the second piece is missing `return {}`, and in
>> fact it's not clear what this code is supposed to do if any if the
>> pointers are null. Even if you did define some sort of behavior in this
>> case, I wouldn't describe this code as something that is easy to understand.
>
>
> Here's how each of those methods could be defined:
>
> Device *Manager::GetDevice(void) { if ( nullptr == this ) return
> {}; return this->pdev; }
> Channel *Device::GetChannel(void) { if ( nullptr == this ) return
> {}; return this->pchannel; }
> Filter *Channel::GetLowPassFilter(void) { if ( nullptr == this )
> return {}; return this->plow; }
> string Filter::GetCodecName(void) { if ( nullptr == this ) return
> {}; return this->codec_name; }
>
> What this means is that you can chain them without checking for a
> nullptr at each step, something like:
>
> pmanager->GetDevice()->GetChannel(0)->GetLowPassFilter()->GetCodecName();
>
> I'm not sure if this is a fantastic idea but I think it's worth looking at.

Right, so you've basically moved the checks for nullptr from the global
GetCodecName to the respective getters. Where's the benefit of your
proposal then?