Bugfix rollup: Fix issue with extra characters in generated SQL; Fix ToAliasReference for already referenced values; Fix Alias/Reference for Maybe Entity #191
Conversation
…ues in mysql. Extra backticks in value reference names
|
@parsonsmatt the code that triggered this bug was unsafeSqlFunction isn't wrapping in parens at all but unsafeSqlBinOp was. Not sure if we should just Never add parens or if this safer check is better. Either way, bumped and were gonna be relying on this commit at work. |
…rence by its new alias source
|
Found another bug while working on this query. Was doing a SubQuery join on a Union and got errors where the union alias ( |
belevy
changed the title
belevy
changed the title
| @@ -767,6 +771,7 @@ instance ( ToAlias a | |||
| type family ToAliasReferenceT a where | |||
| ToAliasReferenceT (SqlExpr (Value a)) = SqlExpr (Value a) | |||
| ToAliasReferenceT (SqlExpr (Entity a)) = SqlExpr (Entity a) | |||
| ToAliasReferenceT (SqlExpr (Maybe (Entity a))) = SqlExpr (Maybe (Entity a)) | |||
There was a problem hiding this comment.
This is a very unrelated question, but why do both ToAliasT and ToAliasReferenceT exist? They seem to do the same thing (?)
There was a problem hiding this comment.
They might be able to be merged but it's not clear to me that they will always be the same. They initially started as associated type families in their respective classes. Semantically they are different though, an alias is something like "value as alias" where an alias reference is like "q.v".
…e general solution than removeTicks
| let | ||
| valueToFunctionArgParens v = | ||
| case v of | ||
| ERaw p f -> first (parensM p) (f info) | ||
| EAliasedValue i _ -> aliasedValueIdentToRawSql i info | ||
| EValueReference i i' -> valueReferenceToRawSql i i' info | ||
| ECompositeKey _ -> throw (CompositeKeyErr SqlFunctionError) | ||
| (argsTLB, argsVals) = | ||
| uncommas' $ map valueToFunctionArgParens $ toArgList arg |
|
@parsonsmatt I fixed #201 while while resolving the conflicts. Seemed better to fix it in one PR and its not much more code. See the commented code for the fix. Additionally removed the other ERaw lambda that was left but it uses the same code that unsafeSqlFunction uses |
in_was generating invalid mysql statements by double wrapping value lists. This existed before 3.3.3.1. Also fixed an issue with the generated value names in subqueries getting "`"s erroneously added for each nestingBefore submitting your PR, check that you've:
After submitting your PR: