You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Problem: Infinite precision integers are far more useful (lowr complexity) than ones limited to a particular size, as users do not have to deal with overflow or underflow.
The F# v1 integers initially used bigint, however since the behaviour of bigint did not extend to the database, it didn't make sense to switch over fully.
To be Evaluated:
We know how to implement this in the Dark engine, but we do not know how to implement this in the database. It's important that Dark values can be used in the database in much the same manner in which they're used in the runtime. This means we need to be able to implement storage and retrieval in the UserDB, as well as queries using the SQLCompiler. We are currently using postgres 9.6, but intend to migrate to a DB supporting live-migrations such as Google Spanner or CockroachDB.
look at our next-gen potential DBs (CockroachDB and Google) and determine if either will support infinite-precision integers
Implementation:
the field of EInt, PInt, and DInt should use a bigint (See Remove bigint #3186)
check any implicit or explicit int conversions (those raise exception in .NET) (see Remove bigint #3186)
the editor handles overflow - this can be removed
determine how to support infinite precision integers in our DB
The text was updated successfully, but these errors were encountered:
pbiggar
changed the title
Infinite precision
Integers should be Infinite precision
Nov 28, 2021
Problem: Infinite precision integers are far more useful (lowr complexity) than ones limited to a particular size, as users do not have to deal with overflow or underflow.
The F# v1 integers initially used bigint, however since the behaviour of bigint did not extend to the database, it didn't make sense to switch over fully.
To be Evaluated:
We know how to implement this in the Dark engine, but we do not know how to implement this in the database. It's important that Dark values can be used in the database in much the same manner in which they're used in the runtime. This means we need to be able to implement storage and retrieval in the UserDB, as well as queries using the SQLCompiler. We are currently using postgres 9.6, but intend to migrate to a DB supporting live-migrations such as Google Spanner or CockroachDB.
Implementation:
bigint
(See Remove bigint #3186)The text was updated successfully, but these errors were encountered: