Fork me on GitHub
#datomic
<
2020-05-01
>
robert-stuttaford13:05:23

what could cause this to happen, @marshall @jaret? no ddb throttling at all, heartbeat totally stable, but all services started getting transaction timeouts, and as you can see on the graph, live index threshold stuck at full. first time we've ever seen this!

jaret13:05:56

@robert-stuttaford would you be able to start a case and share logs, version, config settings? We’d be interested in looking at this in more detail.

robert-stuttaford13:05:22

my colleague Geoff will mail soon!

potetm14:05:33

I’m curious what the answer is when ya’ll find it.

Brian15:05:19

Is it possible to restore a database from a file on disk into an in-memory database? I'd like to write some tests for my code but don't want to point them towards the database on my system since that isn't portable. I know how to restore a database from the command line, what I am looking for is some Clojure code which demonstrates how to take an on-disk backup and restore it into an in-memory Datomic database for use in my tests. Thanks!

kenny16:05:33

Is datomic.query.support a public api?

micah18:05:05

Am pondering the use of :db/fulltext. It’d mean I have to copy data over to new attribute. Is fulltext query really much faster than [(.contains :attr/name ?q)]?

favila19:05:44

that query will always require a full scan of the index

favila19:05:09

fulltext does not. it’s a lucene index with default tokenization and stop words, and you can use lucene queries

favila19:05:39

not sure a “contains” would work reliably because it’s designed for natural language queries and indexing

micah19:05:29

@U09R86PA4 Thanks for the reply. So in short, yeah.. it’ll be faster?

micah19:05:59

.contains kinda worked… here’s how the query ended up

favila19:05:03

it’ll be faster, but depending on what you want, may not be correct

micah19:05:17

well that didn’t copy well

micah19:05:33

'[:find ?e
  :in $ ?q
  :where
  [?e :user/email ?email]
  [?e :user/name ?name]
  [(.toLowerCase ^String ?name) ?name2]
  [(str ?email " " ?name2) ?text]
  [(re-find ?q ?text)]]

micah19:05:00

I see what you mean by correct.

favila19:05:11

yeah that’s an unavoidable full scan of two indexes at once

favila19:05:23

what is the re pattern expected to be?

micah19:05:41

full email, email segment, name segment

favila19:05:22

what do you mean by email segment?

micah20:05:15

Might want to search for “http://cleancoders.com

favila20:05:30

so domain name only?

micah20:05:46

not only… just an example.

favila20:05:51

I’m just wondering if you need the full power of RE or if you just want to match different segments

favila20:05:18

if you want exact matches of different segments, you would be better off creating derived attrs with just those parts canonicalized for indexing purposes, then do exact matches against them

favila20:05:37

that probably wouldn’t work for name

micah20:05:05

Heya… just been browsing http://clubhouse.io and noticed your pm tool uses terms stories and epics, as do I… old school agile XP

micah20:05:17

Do you guys offer trial accounts?

micah20:05:31

Been using Trello but it leaves much to be desired.

favila20:05:39

all accounts up to 10 users are free

favila20:05:56

we’re somewhere on the spectrum between trello and Jira in terms of power and ease of use

micah20:05:42

Sorry… I signed up.. was silly question. Yeah looks cool.

micah20:05:46

Thanks for help