Best place to start contributions

I'm thinking of helping out with this project. Where is a good place to start?

Off-the-cuff, I can see starting to add Postgresql support to the library, but I'm happy to start somewhere less ambitious.

PostgreSQL support would be OK.

Some other things to do:

• Add convenient unpacking support.

  # Something like this
  for a, b in db.rows[:(int64, string)](sql"SELECT 1, 'foo'"):
    ...
  
  # Instead of this
  for row in db.rows(sql"SELECT 1, 'foo'"):
     let a = row[0].i
     let b = row[1].s
     ...
  # or this
  for row in db.instantRows(sql"SELECT 1, 'foo'"):
     let a = row[0, int64]
     let b = row[1, string]
     ...
  

• Prepared statements support.

• Decouple common procs from SQLite-specific procs. This is partly done: procs marked with # Specific are specific to SQLite, whereas procs marked with # Common could be reused by other DMBS wrappers. Perhaps it may be easier to do after adding other wrappers, to make sure that these common procs actutally can be reused.

@xzfc ,

I'm also helping out with the norm project, which is an ORM using Nim's native objects with databases. For that project, adding PostgreSQL is key. So, I'm thinking of this:

1. write a short "How to use ndb" document to complement the technical documentation. I enjoy writing and having such a general document could be useful. And, more importantly, you can check my understanding of the library to make sure it matches up with your vision.

2. Write a first pass at adding PostgreSQL support with a eye towards what parts are specific to sqlite and what parts are general between both databases. On this first pass, I will probably not support array and map columns, but I'll keep them in mind. I'll generate a "[WIP]" PR so that you can see progress and make comments as it goes along. I'll be mostly working on the weekends for this.

Sound good to you?

cc: @moigagoo

Sounds good. Also I'll publish an wiki page about project design choices soon.

Here: https://github.com/xzfc/ndb.nim/wiki/RFC

There is not a way to add commentary to a wiki, not that I can see anyway. So, I'll add my thoughts here.

https://github.com/xzfc/ndb.nim/wiki/RFC#generalize-dbvalue

I've been thinking about this for a bit. I strongly recommend (b) db_* modules should reuse the common DbValue.

A few reasons for this:

• Any app that uses multiple databases will find it very convenient. It's not just about avoiding conflicting namespaces, but also about having a good way to allow conversion. For example, if I have a row I'm importing from postgresql, and I want to save that row on sqlite, and one of the columns is a bool, it would be very nice that sqlite knew what do with a bool. (Which is to convert it to a INT.) And, yes, there are a lot of weird ones that would convert to TEXT/VARCHAR columns in other databases, but I don't see that as a burden, but a benefit that simply needs good documentation.

• I'd keep it in a separate sub-library "x", so that projects that don't even use databases can find it of value. For example, if someone wanted to write a new database library. Essentially, both ndb/sqlite and ndb/postgres would import ndb/x. Speaking of...

• To prep for this possibly becoming part of the standard library, I recommend getting a copy of db_common and making DbValue part of db_common. For now, sqlite and postgresql would import ndb/db_common. But later, they would simply import db_common if/when it becomes part of the standard library.