New diesel features

[weiznich]

After the recent diesel 2.1 release I'm trying to gather some data about what features are considered to be missing by our users. If you have suggestions for different features please leave a comment below.

If you want to support the work on a certain feature via contributions or sponsoring please reach out.

[ethever]

Hey, if cursor-based pagination available out of box, it would great help to using sqlite in my production. https://github.com/SeaQL/sea-orm/blob/master/src/executor/cursor.rs

[Ten0]

Full featured sub-query aliasing:

SELECT users.id, users.username, SUM(p.nb_views) FROM users
INNER JOIN (
    SELECT
        posts.user_id,
        (SELECT COUNT(*) FROM views WHERE views.post_id = posts.id) as nb_views
    FROM posts
) as p ON p.user_id = users.id
GROUP BY users.id

(I'm aware this particular query can be written differently, that's just for the example.)

Notably this would (should) enable batch updates:

UPDATE posts SET content = vals.content
FROM (VALUES (1, "Hello"), (2, "World")) as vals(post_id, content) WHERE vals.post_id = posts.id

[Ten0]

Compile-time check that fields that are not Nullable and don't have a default value are properly set when using insert_into.
NB: Not sure how this interacts with BEFORE INSERT triggers. Maybe there should be an opt-in/out.

[forest1102]

Your project helps a lot.
One feature I want is splitting into multiple schema.rs.
For instance, making the following diesel.toml

[[print_schema]]
file = "db_company/src/schema.rs"
filter = { only_tables = ["^companies$"] }
[[print_schema]]
file = "db_user/src/schema.rs"
filter = { only_tables = ["^users$"] }

Then, the companies schema info is written in db_company/src/schema.rs, while the users schema info goes to db_user/src/schema.rs.

Motivation

• Readability
  • By splitting schema.rs, we can easily search definition of tables.
• Build Performance
  • This feature will parallelize building schema.rs. which will result in better build performance.

Current Behavior

print_schema field only allows only on schema info.
So, what I do right now is select one print_schema and comment out others like the following files.

[print_schema]
file = "db_company/src/schema.rs"
filter = { only_tables = ["^companies$"] }
# [print_schema]
# file = "db_user/src/schema.rs"
# filter = { only_tables = ["^users$"] }

[weiznich]

I've opened a PR that adds building blocks that greatly simplify the implementation of logging and instrumentation libraries on top of diesel. If you are interested in this topic feel free to checkout #3864 and give your feedback there.

[tomcur]

I realize it's not on your priority list, but I was struck by the limited support for views today. (See #43 and #3473.)

[weiznich]

@tomcur That's something we are happy to accept contributions for. At least I do not plan to work on this in the near future. Such contributions could be a complete design and implementation of that feature or just some documentation that explains how you can use a hand written table! macro for this today.