Re: [RFC]mooring API

[Eli Billauer <eli.billauer@gmail.com>]

On 10/08/20 17:39, Oliver Neukum wrote:
>> To me the anchor API is great. If there is unclearances about its API, I
>> >  suppose docs would help. The fact that the URB is unanchored prior to
>> >  calling the completer is intuitive, so there's a clear benefit in that.
>>      
> But is it necessary? DO you ever move URBs between anchors?
>    
I don't think it's relevant what I would or wouldn't do. The question is 
if someone out there would do that. Without any guidelines saying it's 
explicitly disallowed, some will do it. Actually, even if there are such 
guidelines.
>    
>> >  This requires some ungraceful code where almost nobody's looking, but if
>> >  that makes the common programmer's life easier, I think it's a good deal.
>>      
> It would be, if the deal is necessary. In hindsight it still looks
> to me like completion should unanchor an URB.
>    
Once again, I'm not sure that matters so much anymore. The anchor API 
has been around for a while, and there are drivers using it. Adding 
another API will undoubtedly create some confusion. That's why I think 
that the question should be what is functionally flawed with the current 
API, and if that can be fixed.

Regards,
    Eli