Opened 5 years ago
Closed 5 years ago
#32060 closed New feature (fixed)
Expose Random() as a function instead of an expression.
| Reported by: | Nick Pope | Owned by: | Nick Pope |
|---|---|---|---|
| Component: | Database layer (models, ORM) | Version: | dev |
| Severity: | Normal | Keywords: | random, function. |
| Cc: | Triage Stage: | Ready for checkin | |
| Has patch: | yes | Needs documentation: | no |
| Needs tests: | no | Patch needs improvement: | no |
| Easy pickings: | no | UI/UX: | no |
Description
This follows on from #31839 where all other proposals were rejected, but exposing Random() as a function was accepted.
I also propose unifying the random number returned to be in the range [0.0, 1.0) to make it more useful as the backend support differs:
| Backend | Function | Type | Range | Notes |
|---|---|---|---|---|
| PostgreSQL | RANDOM | double precision | [0.0, 1.0) | |
| MySQL | RAND | double precision | [0.0, 1.0) | |
| MariaDB | RAND | double precision | [0.0, 1.0) | |
| Oracle | DBMS_RANDOM.RANDOM | integer | [-2^31, 2^31) | Currently used in Django, but Oracle have deprecated it. |
| Oracle | DBMS_RANDOM.VALUE | number | [0.0, 1.0) | |
| SQLite | RANDOM | integer | [-2^63, 2^63) | We can use Python's random.random() for [0.0, 1.0)
|
This shouldn't adversely affect existing use cases of the expression as it was undocumented and used to support .order_by('?').
We should steer away from setting the random seed as it varies wildly between backends.
We should also avoid any backend-specific arguments for the upper and lower limits for the random number.
For anyone arriving here that wants to generate a random integer from a random number in the range [0.0, 1.0) see the MySQL documentation for a good example.
Change History (4)
comment:1 by , 5 years ago
| Has patch: | set |
|---|
comment:2 by , 5 years ago
| Triage Stage: | Unreviewed → Accepted |
|---|---|
| Type: | Cleanup/optimization → New feature |
comment:3 by , 5 years ago
| Triage Stage: | Accepted → Ready for checkin |
|---|
PR