fix: issue #5613

[krazibit]

No description provided.

[ohjeyong]

This PR cause same error as you said on #5613 on my project which means my PR(#5525) breaks yours and this PR breaks mine. I found the problem.

In PostgresDriver#L606,

buildTableName(tableName: string, schema?: string): string {
    return schema ? `${schema}.${tableName}` : tableName;
}

This is the problem.
Even though the default value of connection option schema is public, if schema option is set on connection option, then buildTableName returns public.some_table_name but if not, then buildTableName returns just some_table_name.

My project doesn't set schema option and your project does.

I think there are two solutions.

Solution 1. Merge this PR and change buildTableName function like below.

buildTableName(tableName: string, schema?: string): string {
    return schema ? `${schema}.${tableName}` : `public.${tableName}`;
}

Solution 2. Change this where statement like this.

WHERE (${tablesCondition})
    AND "column_name" = '${tableColumn.name}'
    AND (
        ("table_schema" || '.' || "table_name") = '${table.name}' OR
        "table_name" = '${table.name}'
    )`;

Solution 1 is much cleaner and better but buildTableName function is referenced a lot, so I don't know is it ok to change like this. @pleerock

Solution 2 is kind of dirty but it will not make any side effects.

[krazibit]

Yeah, you are right @ohjeyong .
I'll add the second solution to the PR

[krazibit]

Or even shorter as a schema variable is available in the method can do

"table_name" = '${table.name.replace(`${schema}.`)}'

[cknuteson-firstlook]

Looking forward to this getting merged. Had to rollback to 0.2.22 for my project

[ohjeyong]

@krazibit Can you add the commit?

[krazibit]

Sure, would add the commit in a bit

[krazibit]

@ohjeyong, can you verify please? thanks

[ohjeyong]

Thanks. LTGM. @pleerock Can you verify please?

[winterec]

This patch fixes it for me

[pleerock]

oh I already merged #5801 before I saw this PR. It looks like author tried to fix same issue, but a bit different way. Is merged approach is correct and I can close this PR? adding @igoraguiar

[krazibit]

closing this as already fixed in PR-5801 as per @pleerock last comment