Add coerceExpr, which coerces between Coercible types
#373
Conversation
|
|
||
|
|
||
| -- | Change the type of an 'Expr' without using @CAST()@, if the | ||
| -- Haskell type can be safely coerced. |
There was a problem hiding this comment.
I'm not sure if it follows that if the Haskell types can be safely coerced then the DB types can be safely coerced. Could you not have a newtype which uses a different DBType?
There was a problem hiding this comment.
I think it's sound only if you ever newtype-derive DBType instances. If that's true then it might not pull its weight: either it'll sound safer than it is, or we'll have to call it something like sometimesButNotAlwaysSafeCoerceExpr. Or use some constraint that indicates that it's specifically a newtype and not just Coercible. (e.g. the one in generic-data.)
There was a problem hiding this comment.
sometimesButNotAlwaysSafe is exactly what unsafe means though. unsafe just means that there is some pre-condition that we couldn't express in the types so you should be careful when using it
There was a problem hiding this comment.
Yeah that's my point. It's slightly less unsafe but still unsafe, which makes me doubt if it's enough of an improvement over unsafeCoerceExpr to include. I might just close this PR.
2 participants
This seems like a good idea (see #175), but I'd want someone who really understands the guts to confirm that.