Re: [DISCUSS] Adding new columnar memory layouts to Arrow (in-memory, IPC, C ABI)

2021-12-15 Thread Wes McKinney
On Wed, Dec 15, 2021 at 6:22 PM Micah Kornfield wrote: >> >> In any case, having memory layouts that support O(# records) >> selections on strings and nested data will greatly benefit some data >> processing systems built on Arrow. > > > Wes, something that still isn't clear to me, are we

Re: [DISCUSS] Adding new columnar memory layouts to Arrow (in-memory, IPC, C ABI)

2021-12-15 Thread Micah Kornfield
> > In any case, having memory layouts that support O(# records) > selections on strings and nested data will greatly benefit some data > processing systems built on Arrow. Wes, something that still isn't clear to me, are we proposing these new encoding for ONLY the C-ABI or do we want to plumb

Re: [DISCUSS] Adding new columnar memory layouts to Arrow (in-memory, IPC, C ABI)

2021-12-15 Thread Wes McKinney
On Wed, Dec 15, 2021 at 3:56 PM Micah Kornfield wrote: > > > > > Big +1 in replacing our current representation of variable-sized arrays by > > the "sequence view". atm I am -0.5 in adding it without removing the > > [Large]Utf8Array / Binary / List, as I see the advantages as sufficiently > >

Re: [DISCUSS] Adding new columnar memory layouts to Arrow (in-memory, IPC, C ABI)

2021-12-15 Thread Micah Kornfield
> > Big +1 in replacing our current representation of variable-sized arrays by > the "sequence view". atm I am -0.5 in adding it without removing the > [Large]Utf8Array / Binary / List, as I see the advantages as sufficiently > large to break compatibility and deprecate the previous

Re: [DISCUSS] Adding new columnar memory layouts to Arrow (in-memory, IPC, C ABI)

2021-12-15 Thread Weston Pace
> I am -0.5 in adding it without removing the > [Large]Utf8Array / Binary / List I'm not sure about dropping List. Is SequenceView semantically equivalent to List / FixedSizeList? In other words, is SequenceView a nested type? The document seems to suggest it is but the use case you described

Re: [DISCUSS] Adding new columnar memory layouts to Arrow (in-memory, IPC, C ABI)

2021-12-15 Thread Jorge Cardoso Leitão
Hi, Thanks a lot for this initiative and the write up. I did a small bench for the sequence view and added a graph to the document for evidence of what Wes is writing wrt to performance of "selection / take / filter". Big +1 in replacing our current representation of variable-sized arrays by

Re: [RESULT][VOTE] Proposed addition to Arrow Flight: Arrow Flight RPC

2021-12-15 Thread David Li
Great. I've merged both your PRs into the new branch, so you can open a PR for the integration tests against that branch next. -David On Wed, Dec 15, 2021, at 13:52, Rafael Telles wrote: > Thanks everyone! > > @David, we prefer to already merge the existing PRs and then add the > integration

Re: [RESULT][VOTE] Proposed addition to Arrow Flight: Arrow Flight RPC

2021-12-15 Thread Rafael Telles
Thanks everyone! @David, we prefer to already merge the existing PRs and then add the integration tests as soon as possible. Em qua., 15 de dez. de 2021 às 11:11, David Li escreveu: > My vote: +1 > > The vote passes with three +1 (binding) votes, one +1 (non binding) vote, > and one -0.5

Re: [RESULT][VOTE] Proposed addition to Arrow Flight: Arrow Flight RPC

2021-12-15 Thread David Li
My vote: +1 The vote passes with three +1 (binding) votes, one +1 (non binding) vote, and one -0.5 (binding) vote. However, we will first merge into a separate branch and implement integration tests before merging into the main branch. JIRA for integration tests: