Nice! Thanks @Hiteshm01

Would you be able to put together a test case for this?

@brianc I found the root cause of it but could only manage a hackish handling in code. Could you please have a look at my latest commit?

@brianc , Were you able to verify it?

Sorry I missed this earlier! Your code looks good to me - there's a merge conflict currently. :( Once you get that resolved I'll happily pull it in! 👍

@brianc , conflict resolved.

Awesome! Thanks! Merging & pushing new patch version now.

This doesn't always work in pg; will produce a test case

edit: it does actually work, it generates INSERT INTO "post" DEFAULT VALUES.

@Hiteshm01 I don't think this behavior is correct - if an empty array is passed, the query should not try to insert anything at all...

At the very least, this requires a major version bump as it causes existing code that did nothing to try and insert data.

@spion , I agree. It should not. But earlier implementation was returning a "select * from table" in this case which was disastrous in any production table. My fix isn't ideal as it is only avoiding trouble.
Suggestions/improvements welcome.

@Hiteshm01 a harmless query like select (1 = 1) would probably be ideal there. Even throwing from node-sql would be better as it would not result with random inserts happening in the DB (which could be even more disastrous - e.g. what if a new AccountState is inserted with the default value of accountId = 1 and money = 0)

In any case, major version (in terms of npm i.e. 0.44) should definitely be bumped.

@spion , That will be a better fix but do you think returning a select statement when requested for an insert statement is correct behavior?

Well, not really - its just backward-compatible behavior. The correct backward-incompatible behavior would be to not generate a query.

Agreed. I will try to implement it over the weekend.

Attempted fix @ #203