On 08/01/2014 10:33 AM, Iñaki Baz Castillo wrote:
> 2014-08-01 9:48 GMT+02:00 Saúl Ibarra Corretgé <[email protected]>:
>> UV_RUN_ONCE doesn't imply not blocking waiting for i/o, that's what
>> UV_RUN_NOWAIT does.
> 
> Any real use case for UV_RUN_ONCE having UV_RUN_NOWAIT?
> 
> I consider UV_RUN_ONCE a bit "problematic" given, exactly, the
> rationale in your previous mail. This is, I consider extremely useless
> blocking until a TCP connection is received, and then exit. No sense
> IMHO.
> 

If you have a timer that kicks in 100ms, UV_RUN_ONCE will block for
100ms. What I showed was an example.

> If there is no real use-cases for UV_RUN_ONCE (that cannot be achieved
> with UV_RUN_DEFAULT and/or UV_RUN_NOWAIT) then I suggest dropping it.
> 

Not going to happen. People embedding libuv into other event loops may
want to iterate the loop at their own pace. UV_RUN_DEFAULT doesn't help
because it will loop for ever, and UV_RUN_NOWAIT doesn't help either
because it always does a zero tiemout poll.

The fact that *you* don't need it doesn't mean other don't have a use
for it.

-- 
Saúl Ibarra Corretgé
bettercallsaghul.com


Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to