(count dump/transit-string)
=> 2806139
(count @conn)
=> 35026

$ gzip -9 ~/Desktop/ds.edn
$ ls -la ~/Desktop/ | grep ds.
... 2966238 Mar 11 02:58 ds.edn
... 1188981 Mar 11 02:56 ds.edn.gz

@mruzekw or do you mean dependency?

is there a benchmark showing overhead of datascript vs hashmap for: * create empty * assoc/update single item * read * small data (i.e. <= 10 kv pairs)

for 10kvs just do it yourself

@misha: what does "kappa" mean? I tried google, but I don't get the reference.

emotion displayed on that dude's face

I don't get it: http://knowyourmeme.com/memes/kappa

if it's intended to convey sarcasm, I don't get the sarcasm in your statement

even for things like 10kvs, since js is gc-ed, I'm not sure how I would measure memory usage

in fact, I think benchmarking datascript vs hashmap would be quite non-trivial since we have to taken into account browser, gc, hotspot, jit

(1) because of gc, it's not clear to me how to measure memory usage; an d(2) because of hotspot/jitting, I feel like any timing would be too noisy

I don't think any of those matter for 10 datoms/kvs, unless it is embedded device, in which case you would not even consider using datascript in the first place

maybe this is a bad idea, but I'm very seriously conidefing replacing many {}'s with datascripts

so it's not just a single datascript with 10 datoms/kvs (though I did not explain this well) -- but lots of places, where I previously used hashmap, I'd now use datascript

because I often use hashmaps to encode "objects"; and if I'm doing that, datascript's query + transact would be better

so I guess I should have rephrased my qeustions as follows:

what's the performance of using 10,000 small datascripts vs 10,000 small hashmaps

that's the question I should have asked in retrospect

single db instead of lots of maps? or lots of dbs instead of lots of maps?

lots of dbs instead of lots of maps

I'm curious how "lightweight" datascript is

I doubt benchmarking single 10datom db vs 10 kvs would give you any useful info for the 10k tiny dbs

yeah, I should have rephrased the question as: 10,000 small datascript dbs vs 10,000 small hashmaps

(time
  (dotimes [_ 10000]
     (hash-map [:foo :bar])))   
"Elapsed time: 6.585000 msecs"

(time
  (dotimes [_ 10000]
     (ds/transact! (ds/create-conn) [{:foo :bar}])))   
"Elapsed time: 1248.580000 msecs"

(1) you win

(2) it's 200 times slower?

I don't know, but that would give me enough info to re-consider replacing every {} with a tiny db

it's enough into to decide against

this is what I meant by "for 10kvs just do it yourself"

these micro benchmarks are the worst

you gain no knowledge through them, they more often just mislead you

@mruzekw When I first added datascript it added ~30kb gzipped to the project. 15kb gzipped if you don't need the query engine (which I don't use and stripped).

Thanks @rauh