UPDATED: sqlc.narg() - allow user-specified nullability#1536
UPDATED: sqlc.narg() - allow user-specified nullability#1536kyleconroy merged 12 commits intosqlc-dev:mainfrom
Conversation
| Number: argn, | ||
| Location: fun.Location, | ||
| }) | ||
| } |
There was a problem hiding this comment.
aside: encapsulating the positional tracking into named.ParamSet significantly cleaned up this code.
skabbes
changed the title
| @@ -54,25 +54,31 @@ func Mutate(raw string, a []Edit) (string, error) { | |||
| if len(a) == 0 { | |||
There was a problem hiding this comment.
I was having a bit of trouble with this, so I tested the heck out of it and these were the resulting edits.
There was a problem hiding this comment.
The edit code is super hacky so I'm not surprised you needed to add all these tests. Hopeful in the future we can get rid of all of this code and generate SQL using an AST.
|
@kyleconroy I went ahead and implemented |
| @@ -54,25 +54,31 @@ func Mutate(raw string, a []Edit) (string, error) { | |||
| if len(a) == 0 { | |||
There was a problem hiding this comment.
The edit code is super hacky so I'm not surprised you needed to add all these tests. Hopeful in the future we can get rid of all of this code and generate SQL using an AST.
|
It's a lot to ask, but can you also add docs to https://docs.sqlc.dev/en/stable/howto/named_parameters.html in a separate PR? Want to make sure that people know this option exists. |
|
ah, I didn't look hard enough for the docs. But yes I can do that. |
|
Added here: #1579 |
|
Just kidding, one that actually passes CI here: #1580 |
2 participants
EDITED from original PR after discussion here: #1525
Exec Summary
I implemented user-specified nullability with
sqlc.nargDetails
Most of the work here came from building types to represent set of named parameters, which tracks the name and nullability status of any parameter (inferred or user defined).
named.ParamThis type can track its nullability from the sqlc engine (inferred nullability) separately from any user-specified nullability, and always prefers the user-specified version to the inferred one.
named.ParamSetThis type is essentially a collection of params, and helps build up parameters, and encapsulate their positional representations as well. As you can see - this type significantly cleaned up
compiler/resolve.goNext Steps
This code works, is tested. I think a code review would be great, and maybe a pointer to how to update the docs (or maybe the doc repos are still private).