Fork me on GitHub
Slack Archive

This page is not created by, affiliated with, or supported by Slack Technologies, Inc.

2022-08-15

Channels

#datomic
<
2022-08-15
>
robert-stuttaford11:08:16

has any work been done to make Datomic on-prem's metrics and logging compatible with OpenTelemetry? we would love to use http://Honeycomb.io but not having the database in the picture is a pretty big blocker 😅

💯 3
Drew Verlee15:08:38

I think of an index as a way group a set of things by some property, given a uuid is about randomness, it's unclear to me when a https://docs.datomic.com/on-prem/schema/identity.html#squuids aka a uuid with an ordered time component, would be useful. except for a general query against all such uuids? like give me everything in the db from the last week? It's really indexed by just the time aspect right?

favila15:08:37

The point of a squuid is to reduce garbage from index fragmentation, full stop

👀 1
favila15:08:23

new values accumulate on the “tail” of the tree vs randomly distributed throughout the entire tree

👀 1
Drew Verlee15:08:08

So it will improve performance in general query cases, not just the time based one i was thinking of?

favila15:08:21

it won’t improve performance in queries, except in pretty specific cases

👀 1
favila15:08:20

it will reduce the amount of garbage segments the transactor produces when it makes an index, and may make preparing the new indexes faster also.

👀 1
👍 1
Drew Verlee18:08:54

What type is a db/id? can it be a long id? i'm trying to make sense of this https://github.com/tonsky/datascript/issues/292 about datascript, i'm not sure that they mean and i'm worried im missing something. by jvm do they mean datomic? why would the jvm have any say on anything "id" related? > It seems that long ids are no longer supported on JVM since 0.18.0. I also added a validation that throws on both JVM and JS if value is out of range.

favila19:08:43

The comment means that datascript rejects entity ids in the long range (specifically > 0x7FFFFFFF)

👍 1
favila19:08:06

it rejects them on the JVM even though they would be fine, because they would not be fine on js/cljs

Drew Verlee19:08:38

yep, that makes sense and mirrors what i'm seeing on my end to. datomic says an eid should be a nat-int? and that seems to include the eids that are causing exceptions in datascript. Thanks for the help :thumbsup: