Sqlc: Data access simplified. Throw away your ORM | Hacker News Reader

Sqlc: Data access simplified. Throw away your ORM

docs.sqlc.dev

24 pointskaashmonee22 comments

A great option for Go projects where you aren't exploring your queries as changing and regenerating is cumbersome.

Also, SQLC doesn't allow for dynamic query building so think about an input search struct from which you may add where clauses — or not.

Also doesn't support multiple inserts with N rows being inserted.

Otherwise for standard queries, it's great.

Static templates for SQL becomes a chore to express dynamic queries. A dynamic query depends from incoming request. For example search filters should add or delete conditional expressions for a query.

I have a little research[1] on dynamic queries.

[1]: https://github.com/baverman/sqlbind?tab=readme-ov-file#dynam...

If you're writing a lot of dynamic queries I think you should opt for a query builder, not an orm.

For me, I write almost nothing but static queries, SQLc is just so much nicer to use.

I don't mind having to do the odd dynamic query from scratch.

I just wish sqlc supported named Params in MySQL, the resulting function param ordering is a little annoying.

The issue with templates they could handle only simple cases or very specific complex static queries. Simple ones are a way easier with ORM/query builder. Highly depends from ORM though.

And a big NO-NO for me templates force you to repeat the same SQL in many slightly different queries. SQL composability is not a thing with templates.

Simple queries aren't any easier with ORM, that's the point of sqlc, you write the SQL and then call it as a function. There's no extra abstraction or added steps on top.

Repeating portions of sql isnt really an issue for me either, it's the business end of the database, you want it all there in one location, spreading it around in a composible fashion is just one abstraction too far. I know SQL, I don't want to learn anything else on top.

I guess you talk from only a Go perspective where you don't know any good ORM libraries. Other languages have quite convenient ORMs. Some allows even auto migrations, creating migration SQL for your.

I did a quick look at GORM. And it's a way easier and safer to use for simple cases.

Numbered parameters become fragile quite fast for even simple queries. Basically you couldn't safely use query until carefully revising a template. A template change could potentially break existing queries. It's kinda scary.

If library supports named parameters they should be used by default in examples.

Its probably to be consistent across each platform, MySQL doesn't support named queries so they are likely avoiding using them for that reason, either that or the docs are old. :)

This is cool, but most ORMs have support for raw SQL.

We have achieved at a point were we call SQL "raw". Am I this old?

I've been hearing people calling SQL "raw" for the past 10 years so... probably?

It's only called "raw" when called from an ORM, as opposed to generated SQL.

It became raw since query builders started to try to mimic SQL. Quite long ago.

SQLc isn't just raw SQL.

SQLc you write the queries, it generates the boilerplate functions to execute them.

This works better than an ORM because you don't have to deal with an ORM.

How SQLc would solve following issue.

For example I have original query:

    SELECT * from users where following_count > $1 and followers_count < $2;

Then some refactoring later it becomes:

    SELECT * from users where enabled and followers_count < $1 and following_count > $2;

As I understand go API would not change it still query(int, int).

Use hql named queries with named parameters

What's hql? If it's something related to hybernate than your comment is slightly out of context.

Hibernate Query Language. How is it out of context?

Use Named Native Query and you have SQL queries.

I really don’t understand the objection to ORMs. By that logic, you might as well reserve a block of memory and offsets instead of class/struct attributes.

I did not object against ORMs. Please reread thread. I also have issues with SQLc, and please note it's Go library. Java is out of context.

Is the question about how to version the query? You have comments before the query that allow you to name the exported function so you could have both queries and name them something different.

I'm not sure that's an issue? It would be an odd refactor to add a new side effect to the query you didn't want to apply across the board.

More likely you would introduce a new query that would get a new function call.

usersWithCountBetween(a, b)

enabledUsersWithCountBetween(a,b)

If you mean, how to handle the addition of a new enabled flag that is passed in, that too would either require refactoring or a new function.

These aren't really things I would consider a problem but maybe I'm missing something?

Oh and you can configure the number of params before it replaces the params with an interface so things like this are easier to manage over time.

Reply on news.ycombinator.com