is there any reason datascript tends to use attributes o the form: :foo/name :dog/age :student/gender instead of "name", "age", or "gender" ?

That's not really related to datascript, it's generally recommended to switch to namespaced keywords.

But it does fit the "entity" concept nicely imo.

Initially it was recommended by Datomic because it leads to better index locality

oh right, if I have :student/name and :computer/name and :employee/name

E.g. person names would not mix with goods names

Yes

then when I lolk up :computer/name, it's not returning stuff with :employee/name or :student/name

Turned out it leads to better code as well

You see a piece of code anywhere and understand immediately with what it works: persons, goods, etc

Also much easier to search and replace if you need to rename/refactor/adjust something

It's impossible to search for just "name", too many unnecessary stuff will pop up as well

@tonsky: I completely agree with your logic. Why does this logic not also apply to vec2? i.e. {:x 20 :y 40} vs {:vec2/x 20 :vec2/y 40} people tend ot use the former more than the latter

Because x and y are not domain names

Is it still the case that I can't delete non-existing refs? I am looking at the code for avoiding this and I'd really like to get rid of it. Apart from seeming unnecessary, it slows down my app considerably, because when doing mass-deletes I have to first find all existing entities, just to avoid deleting non-existing ones.

(let [c (ds/create-conn)]
  (ds/transact! c [[:db.fn/retractEntity 100]]))
=>
{:db-before {},
 :db-after {},
 :tx-data [],
 :tempids {:db/current-tx 536870913},
 :tx-meta nil}

(let [c (ds/create-conn)]
  (ds/transact! c [[:db/retract 100 :foo :bar]]))
=>
{:db-before {},
 :db-after {},
 :tx-data [],
 :tempids {:db/current-tx 536870913},
 :tx-meta nil}

Hmm. It seems it is possible now — I wonder when that changed. Digging into my code now.

what do you mean by deleting non-existing entities?