For CRUD stuff I have found datomic to be hard to maintain if your domain demands lots of attributes (even if trying to be as generic as possible), you need discipline (good docs) and lots of application code for checks and to force constraints. In sql/rdbms world you have all the atrributes nicely organized(within tables) with good check/constraints available at hand (waits for easy vs simple comment).
With datomic you get fleixble schema but at a high cost IMO.
Mongo moves your schema entirely to the application, in datomic you have to define your attributes (columns) before hand (enabling some constraints, like uniqueness and value types) but its up to the app to group them however you want (no tables), they called it the universal schema, it also enforces acids semantics. So is not as bad as mongo for structured and relational data.
8 comments
[ 3.3 ms ] story [ 20.9 ms ] threadAny guidance on wire protocol (BOLT, TinkerPop, some custom HTTP, ...?) + early users? Guessing ThreatGrid..
http://blog.cognitect.com/cognicast-transcripts/114
With datomic you get fleixble schema but at a high cost IMO.
You could say the same thing about mongo or any other nosql document-db...