std.experimental.logger formal review round 3

[via Digitalmars-d]

On Sunday, 5 October 2014 at 23:40:32 UTC, Kevin Lamonte wrote:
> On Sunday, 5 October 2014 at 17:06:06 UTC, Sean Kelly wrote:
>
>> Having an identifier for logging is a bit different.  Would 
>> using the MessageBox address be sufficient?  I'd be happy to 
>> add a Tid.id property that returns a value like this.  I'd 
>> rather not try to generate a globally unique identifier though 
>> (that would probably mean a UUID, which is long and expensive 
>> to generate).
>
> I think Tid.id returning the MessageBox address would be fine 
> for logging purposes.  The main value is being able to 
> distinguish messages coming in at the same time from multiple 
> threads.  Even if a MessageBox address was re-used by a new 
> receive()er after a previous one exited I doubt it would 
> confuse users very much.  I thing that a doc on Tid.id like 
> "this value will not be the same as any other Tid currently 
> existing, but might be the same as a Tid that has exited 
> previously" would be sufficient.

A moving GC can affect the address, too, but for your purpose it 
would still be fine, you just mustn't put too much significance 
into the actual value.