3 ms·

This is great. I built something a bit lower-level than this targeting MySQL not long ago. This library has taught me a couple language features I didn't know,

This is great. I built something a bit lower-level than this targeting MySQL not long ago. This library has taught me a couple language features I didn't know, however. The SQL tag is pretty clever. I was pleasantly surprised to see transaction support (I feel like people who don't actually use their libraries in real products tend to leave this kind of thing out). I noticed the support for soft-delete (which seems to be a simpler thing in PgSql than in other relational db systems), which is also nice. I think another fairly easy, generic win would be a way to specify that a model has audit tracking fields (createdAt/By, modifiedAt/By, deletedAt/By, etc.). I know some database servers also have a way to track this separately (no idea whether PgSql specifically does), but there are use cases for showing that kind of stuff at an application level as well (and also can make ETL type jobs a bit less painful). All in all, great work. It looks very polished!

Seb-C 6y ago
Thanks! It was a bit tricky to solve the transactions problem, because I did not want to abstract the BEGIN / COMMIT / ROLLBACK instructions, but at the same time I needed to provide something to ensure the integrity of a transaction made of multiple commands. As for the audit fields, I decided not to include it because it very simple to implement, and would be clearer to do it specifically. Most of those fields could be implemented by just adding a default value in the right function.
- twodave 6y ago
  Another interesting way to solve the transaction problem is following more of a Unit of Work pattern. Then the transaction is more or less held until the unit is committed. Or rolled back. This is also wonderful for writing integration tests because your repositories all contribute to the same UoW and your test can just roll back when it’s finished to return to a clean state.

Seb-C 6y ago

Thanks! It was a bit tricky to solve the transactions problem, because I did not want to abstract the BEGIN / COMMIT / ROLLBACK instructions, but at the same time I needed to provide something to ensure the integrity of a transaction made of multiple commands. As for the audit fields, I decided not to include it because it very simple to implement, and would be clearer to do it specifically. Most of those fields could be implemented by just adding a default value in the right function.

twodave 6y ago
Another interesting way to solve the transaction problem is following more of a Unit of Work pattern. Then the transaction is more or less held until the unit is committed. Or rolled back. This is also wonderful for writing integration tests because your repositories all contribute to the same UoW and your test can just roll back when it’s finished to return to a clean state.