On Wednesday, September 2, 2015 at 3:32:55 AM UTC-7, Tiago Cardoso wrote:
>
> Hi,
>
> I've been designing this framework to perform certain tasks on distributed 
> devices by ssh. I've been using celluloid (and celluloid-io) to leverage 
> the concurrency for a mixed of thread-based/event-based. It has so far been 
> working well, except on a certain use case.
>
> Part of those tasks sometimes involve communicating with a database to 
> retrieve some information to send to the device, for which we've been using 
> Sequel. 
>
> It has been working, but it was noticed that sometimes the query takes too 
> long, and it blocks the celluloid-io internal reactor. This is because 
> sequel uses the ruby TCPSocket (and eventually, raw IO.select calls), and 
> these calls are not handled by celluloid-io. 
>
> Celluloid IO provides its own TCP/UDPSocket duck types, and here (
> https://github.com/celluloid/celluloid-io/wiki/Using-Celluloid%3A%3AIO-with-existing-Ruby-libraries)
>  
> it proposes an API to "inject" these to be used by the clients. Now, I 
> don't know how to continue this debate, as this probably is being handled 
> internally in the db client adapters, and I don't know if these already 
> provide some kind of API for this, and its level of support for such 
> event-based architectures (I know that event-machine had to develop its own 
> adapters at some point). So, shall we continue the debate here, or should I 
> need to discuss this with every db client adapter maintainer?
>
> Many thx in advance,
> Tiago
>

Yes, this is a driver issue, not a Sequel issue, as Sequel doesn't use 
sockets directly.  So you'd need to talk to the driver authors about it.

Thanks,
Jeremy 

-- 
You received this message because you are subscribed to the Google Groups 
"sequel-talk" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sequel-talk.
For more options, visit https://groups.google.com/d/optout.

Reply via email to