Allow for mixing parameter styles

[ovadbar]

This should allow for mixed styles of parameters.

[kyleconroy]

Thanks for the patch! I agree that it doesn't make much sense to require a single parameter style. A few request:

• please pick a more descriptive name for the tests, such as sqlc_arg_mixed or params_mixed.
• Move the current tests into a postgresql subdirectory and add the same test for MySQL

[ovadbar]

Thanks for the feedback.

(One thing I did notice is if you have a query as follows, the numbers always precede the sqlc.args convetion)

SELECT count(1) FROM table_name WHERE id > $1 AND id < sqlc.arg(id);

[dant4301]

This would be great fix and would be awesome if merged.

There is an outstanding issue w.r.t using named_parameter for LIMIT for MySQL. We can use this fix to sort of work around it in the short term, but for the below:

I applied the patch and tried it for the query:

SELECT * FROM books WHERE book_id >= sqlc.arg(min_book_id) ORDER BY title ASC LIMIT ?;

The generated code looks like this:

type GetBooksByNameParams struct {
        Limit                 int32 `json:"limit"`
        MinBookID int64 `json:"min_book_id"`
}

func (q *Queries) GetBooksByName(ctx context.Context, arg GetBooksByNameParams) ([]Category, error) {
        rows, err := q.query(ctx, q.getBooksByNameStmt, getCategoriesByName, arg.Limit, arg.MinBookID)
...

Aren't the orders of parameters arg.Limit and arg.MinBookID switched?

[ovadbar]

Yeah I kind of realized that was an issue but, with postgresql you can use dollar numbers format so it is not an issue and I did not think of how it would work with mysql.

I can look into fixing it but I really do not have the time.

[kyleconroy]

I can look into fixing it but I really do not have the time.

Understandable. I also don't have the time to look into fixing this. @dant4301 If you'd like this feature, you're going to have to get your hands dirty and take a shot at fixing this PR. I'm going to close this for now, but feel free to fork this branch and open up another PR if you can get it to work.

[ovadbar]

@dant4301 can you look at #1071 and see if the new pull request solves your problems

[dant4301]

@ovadbar, I do not use sqlc now since I did a limited amount of codegen just enough for my needs. Thanks again though.