Hi Atamert,
воскресенье, 17 мая 2015 г., 19:35:57 UTC+6 пользователь Atamert Ölçgen
написал:
>
> I’m new to Clojure async operations (while have a good understanding of
>> other things) and want to get a bit of advice. Atoms & agents still confuse
>> me.
>>
>> What I’m implementing is a small REST webservice with custom in-memory
>> database. Database is indexed by unique key, basically it’s a map (I’ll
>> denote it as MAP). There are two basic operations:
>>
>> *PUT /api/…/.* Initiate an update for given KEY. The HTTP response
>> should be returned immediately, and the processing should be done
>> asynchronously. The processing consists of downloading a data file from web
>> and doing some data crunching, the result should go into MAP.
>>
>> *GET /api/…/.* Use (get MAP KEY) and request params to generate
>> response synchronously.
>>
>> The tricky part (well, for me :) is that parallel PUT requests should be
>> possible. The downloading and processing can and should go in parallel,
>> independently, whereas updating the MAP should come synchronized,
>> one-after-the-other.
>>
>> How would you implement this?
>>
>
> While this can be done with atoms & agents, perhaps you should check out
> core.async or Lamina first.
>
> Neither atoms nor agents can be updated in a parallel manner. Only one
> update runs on a single atom/agent at a given time. If you really need
> parallelism and the updates affect small parts of MAP, perhaps you can
> model it using refs, but then it's not one big map anymore. (Still you
> can't update a single ref in more than one thread at the same time.)
>
> You probably also want to control how the files are downloaded and
> processed, so that there's not duplication of effort. So an event driven
> solution seems to fit this problem.
>
In fact, I don't use the MAP as a whole, each read/update operation is
scoped by KEY. So a map of refs sounds good.
Speaking of event driven solution and core.async. I came up with the
following scheme:
[PUT request handler] > processing queue channel > Download/process
> update queue channel > [updating the MAP].
*[updating the MAP]* is basically a (go ...) block which does synchronous
updates.
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
