Seems like a good way forward, even if we need to iron out the details.
We will need a good compression for the delete vectors if we would like to
store then in the metadata files.

+1 for adding an initial draft

Amogh Jahagirdar <[email protected]> ezt írta (időpont: 2026. jún. 9., K,
2:47):

> I had a comment about the exception offsets, but +1 (binding) from me.
>
> Thanks,
> Amogh Jahagirdar
>
> On Mon, Jun 8, 2026 at 6:08 PM Micah Kornfield <[email protected]>
> wrote:
>
>> +1 (non-binding for me).
>>
>> I left some comments that I think might help with clarity for
>> implementers.
>>
>> Ryan, given the limited nature of the vote, what do you see as the
>> decision factors for actually using this?
>>
>> Cheers,
>> Micah
>>
>> On Mon, Jun 8, 2026 at 4:21 PM Ryan Blue <[email protected]> wrote:
>>
>>> +1 from me
>>>
>>> On Thu, Jun 4, 2026 at 9:01 AM Ryan Blue <[email protected]> wrote:
>>>
>>>> Hi everyone,
>>>>
>>>> I mentioned in the thread on adding a bitmap format for small bitmaps
>>>> that I think the next steps are to get a draft spec into git (PR #16518
>>>> <https://github.com/apache/iceberg/pull/16518>)  and then iterate on
>>>> it and implementations. That will give us more information to move forward
>>>> with the Mumbling bitmap format in v4 metadata or not.
>>>>
>>>> This vote is to add the spec to version control. It does not mean that
>>>> we are using the format in tables or other places. Those uses will be
>>>> covered by future votes.
>>>>
>>>> Please vote in the next 72 hours:
>>>>
>>>> [ ] +1: Add the Mumbling bitmap spec
>>>> [ ] +0: . . .
>>>> [ ] -1: Do not add the spec to git because . . .
>>>>
>>>> Thanks,
>>>>
>>>> Ryan
>>>>
>>>

Reply via email to