Mon Mar 17 10:28:37 CET 2014
On Sat, Mar 15, 2014 at 6:54 PM, Loïc Hoguin <essen@REDACTED> wrote:
> In what release port monitoring would land, presumably?
I don't know. If someone wants it the fastest way to get it in is
submitting a patch. Doing the coding should be relatively similar to how
link works with ports, so shouldn't (tm) be too hard.
> Speaking of erlang:monitor, we have also been talking about adding a
>> erlang:monitor/3 which would as a third argument take the identifier
>> that you want to be sent back to the caller instead of the current
>> reference. Is that something which could be useful to anyone?
> That does sound like something I could use, yes.
> The Gun HTTP client API (a few days from being public) creates a process
> per connection, and then each request on the connection gets a reference.
> If the user wants to do many requests on the same connection, then creating
> a separate reference for the monitor makes sense, however I could use what
> you propose when the user wants to do a single request and then close the
> connection, we can reuse the reference from the request in the monitor
> instead of creating two references.
> Doing it the other way (use the monitor ref as the request ref) would
> complicate the API, so that's not an option.
Good to know!
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the erlang-questions