This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2021-05-22
Channels
- # announcements (1)
- # aws (11)
- # babashka (10)
- # beginners (49)
- # calva (32)
- # cljsrn (3)
- # clojure (123)
- # clojure-australia (1)
- # clojure-dev (2)
- # clojure-europe (27)
- # clojure-spec (1)
- # clojurescript (22)
- # clojutre (1)
- # code-reviews (1)
- # conjure (3)
- # editors (32)
- # emacs (3)
- # graalvm (12)
- # kaocha (1)
- # lsp (1)
- # malli (2)
- # off-topic (1)
- # schema (2)
- # shadow-cljs (32)
- # spacemacs (6)
- # tools-deps (4)
For those interested in dead simple databases – I’ve made a https://youtu.be/t0gG1zzIT_c for https://github.com/ribelo/doxa by @huxley https://github.com/spacegangster/rf-doxa-etude Easiest diffing and syncing I’ve ever seen.
this looks exciting. also the first sentence reminds me of https://imgs.xkcd.com/comics/standards.png : D
when people say datascript is slow, I would like to see the instances it is slow, because when I used it, it was possible to optimize it quite well, and would be interesting where the different experience comes from, is it a different part of the API, is it the scale?
I measured it about 2 years ago - compared a simple datascript
setup with a simple custom solution similar to doxa
.
datascript
ended up being ~10 times slower, which was definitely noticeable on large tables that I usually deal with.
Hehehe, but pluralism is our strength! Is it easy to diff and patch in Datascript? Doxa is about balance, as I got it.
I take 15 standards over a single standard any time 🙂 Better to have something to choose from 😄
@U0VQ4N5EE Doxa was not intended to be faster than datascript, because that's not its main problem for me. The fact that it is faster is a coincidence. And how slow datascript
is, let's see datalevin
, which despite the fact that it stores data on disk is several times faster than datascript
which keeps data in memory.
@U2FRKM4TW doxa uses timbre, editscript and meander, so it may punch well above its perceived weight. We’re about to find out
@U2FRKM4TW while I trust you fully, note that this is still the same, we talk about datascript being slow without me seeing the instance when it's slow : )
https://github.com/juji-io/datalevin/raw/master/bench/datalevin-bench-query-01-05-2021.png
ok, can I see the code that generated the chart please? and yes, the talk about performance here is 100% independent of doxa and doxa's promises, I do realize and I do intend to try doxa out when I will have a bit more time
https://github.com/juji-io/datalevin/tree/master/bench/src-datascript/datascript_bench found it
I am not a senior developer with 20 years of experience who knows exactly what he is doing.
only people who did the same thing for 20 years continuously know what they are doing 😄 people who do new things, experience only teaches how little we know.
although I didn't quite hit the 20 years mark, maybe it will change in a couple years : D
How fast do you need to be though? For Lightpad occasional ~250ms queries over 10k-100k elements would be completely fine. But I’ve spent weeks writing and testing my own diff accumulation / patching solution.