meander

All things about https://github.com/noprompt/meander Need help and no one responded? Feel free to ping @U5K8NTHEZ
grounded_sage 2020-02-19T13:54:18.287200Z

How would I capture 2 pieces of information and join them inside the match. Eg: !date and !time are captured and I can create a datetime with whatever lib I choose.

(m/rewrites {:venue venue-data
             :event (group-by :related-id event-data)}
            
            ;; Pattern
            {:venue (m/scan {:related-id ?id
                             :location (m/app str/trim ?location)})
             :event  {?id [{:date (m/pred between-x-y? !date)
                            :start-time !start-time
                            :attendance (m/pred min-size? !event-size)
                            
                            }...]}}
            ;; Action
            {:id ?id
             :name ?location
             :location [{:event-name !event-name
                         :date-time !date-time
                         :event-size !event-size} ...]})

grounded_sage 2020-02-19T14:32:31.287500Z

Oh I can apply (m/app ….) on the action.

1👍
grounded_sage 2020-02-19T14:32:37.287700Z

Super easy 🙂

grounded_sage 2020-02-19T19:50:15.294200Z

Hit memory limits today on joining 2 datasets that weren’t particularly that large. Maybe like 50,000 and 30,000 Clojure maps with little in them. Nothing to criticise though. I think this task is better left for a database. Maybe using meander inside a transducer and process them one at a time and write to db in a stream.

jimmy 2020-02-19T19:51:55.295500Z

Did you hit a memory limit because of meander? Or because your dataset is just too large for your heap size?

jimmy 2020-02-19T20:36:37.299900Z

If you use something like visualvm or mission control you can inspect the memory of your application. You'd be surprised at how much memory maps can take up.

grounded_sage 2020-02-19T20:46:27.300700Z

Yea and it probably holds onto the parsed CSV before it was converted to EDN because of immutability?

grounded_sage 2020-02-19T20:47:23.301400Z

Actually I don’t know how I got those numbers I said earlier. It’s was more data than that :rolling_on_the_floor_laughing:

grounded_sage 2020-02-19T20:47:45.301700Z

Must have been the other files I joined on

grounded_sage 2020-02-19T20:50:12.302400Z

I’m thinking of bringing in datahike to write to disk and doing the transaction during the action step.