Strange array representation before parsing

[btd]

Hi there.

I have several arrray of types _int4 and _int8. When updating type parser for _int8 i noticed that db returns for this column such repesentation: [0:3]={2,4,3,1}. Have you seen something like this?

I am using AWS RDS 10.4

[btd]

I found this comment in pgjdbc driver:

      // Starting with 8.0 non-standard (beginning index
      // isn't 1) bounds the dimensions are returned in the
      // data formatted like so "[0:3]={0,1,2,3,4}".
      // Older versions simply do not return the bounds.
      //
      // Right now we ignore these bounds, but we could
      // consider allowing these index values to be used
      // even though the JDBC spec says 1 is the first
      // index. I'm not sure what a client would like
      // to see, so we just retain the old behavior.

[charmander]

See https://github.com/bendrucker/postgres-array/issues/2.

[btd]

@charmander thanks. I missed this issue. But anyway i prepared PR for postgres-array. It works fine for my workload

[bendrucker]

Hi, I think we've fully discussed this issue in bendrucker/postgres-array#2. I'll leave comments on your PR. I'm glad it works for you but I don't think changing the default behavior is something we'll accept given my existing objections in the issue.

[bendrucker]

Good find on pgjbdc:

Right now we ignore these bounds, but we could consider allowing these index values to be used... I'm not sure what a client would like to see, so we just retain the old behavior.

That says it all to me. It's ambiguous, they already did it, and so now it stays. If you come from pgjbdc to any other client without this same behavior, your code breaks. If pgjbdc didn't make an assumption for you in the first place, you'd discover the mistake in your query/data right away and fix it.