Hi, Thanks for the discussions.

a) What's the reasoning behind merging the DV and position delete record
counts into `position_delete_record_count`? Would it be better to maintain
a separate counter? This way, if position delete files are deprecated in
the future, DVs remain unaffected.

Make delete counts required to avoid ambiguity w.r.t NULL vs unknown.

b) While we are at it, should we also handle `total_record_count` in the
same manner as we decided to keep all counters initialized as zero if not
computed during implementation.

Thanks,
- Ajantha


On Mon, Feb 3, 2025 at 2:35 PM ConradJam <czy...@apache.org> wrote:

> +1 (non-binding)
>
> Steven Wu <stevenz...@gmail.com> 于2025年2月3日周一 14:08写道:
>
>> +1
>>
>> The spec change makes sense. left a question in the PR.
>>
>> On Sun, Feb 2, 2025 at 8:52 PM roryqi <ror...@apache.org> wrote:
>>
>>> +1
>>>
>>> Amogh Jahagirdar <2am...@gmail.com> 于2025年2月2日周日 10:16写道:
>>>
>>>> +1
>>>>
>>>> On Sat, Feb 1, 2025 at 11:05 AM huaxin gao <huaxin.ga...@gmail.com>
>>>> wrote:
>>>>
>>>>> +1 (non-binding)
>>>>>
>>>>> On Sat, Feb 1, 2025 at 8:50 AM Manish Malhotra <
>>>>> manish.malhotra.w...@gmail.com> wrote:
>>>>>
>>>>>> +1(nonbinding)
>>>>>>
>>>>>> On Sat, Feb 1, 2025 at 2:49 AM Russell Spitzer <
>>>>>> russell.spit...@gmail.com> wrote:
>>>>>>
>>>>>>> +1
>>>>>>>
>>>>>>> On Sat, Feb 1, 2025 at 3:01 AM Anton Okolnychyi <
>>>>>>> aokolnyc...@gmail.com> wrote:
>>>>>>>
>>>>>>>> Hi all,
>>>>>>>>
>>>>>>>> I propose the following updates to our partition stats spec in V3:
>>>>>>>>
>>>>>>>> - Modify `position_delete_record_count` to include a sum of
>>>>>>>> position deletes across position delete files and DVs
>>>>>>>> - Keep `position_delete_file_count` to represent the number of
>>>>>>>> position delete files (ignoring DVs)
>>>>>>>> - Add `dv_count` to represent the number of DVs
>>>>>>>> - Make delete counts required to avoid ambiguity w.r.t NULL vs
>>>>>>>> unknown.
>>>>>>>>
>>>>>>>> Here is the PR with the spec update:
>>>>>>>> https://github.com/apache/iceberg/pull/12098
>>>>>>>>
>>>>>>>> - Anton
>>>>>>>>
>>>>>>>

Reply via email to