Inconsistent query table results with/without select columns

[bdpedigo]

I am getting inconsistent results in a CAVEclient query to materialization engine when I do/do not use the select columns feature

MRE:

import caveclient as cc
import numpy as np
import pandas as pd

client = cc.CAVEclient("minnie65_phase3_v1")

timestamp = pd.to_datetime("2021-07-01 00:00:00", utc=True)

soma_ids = [
    292864,
    291116,
    303149,
    264824,
    292670,
    260541,
    301085,
    294825,
    292649,
    298937,
    262678,
]

old_roots = client.materialize.query_table(
    "nucleus_detection_v0",
    filter_in_dict={"id": soma_ids},
    timestamp=timestamp,
).set_index("id")["pt_root_id"]
print(old_roots)

old_roots_w_select = client.materialize.query_table(
    "nucleus_detection_v0",
    filter_in_dict={"id": soma_ids},
    select_columns=["id", "pt_root_id"],
    timestamp=timestamp,
).set_index("id")["pt_root_id"]
print(old_roots_w_select)

Output

id
291116    864691135526260315
262678    864691136602049489
292864    864691135941381492
298937    864691135771983611
292670    864691135568038892
294825    864691135618041743
292649    864691135394228469
260541    864691136109120184
301085    864691135776863584
264824    864691136423000111
303149    864691135132830624
Name: pt_root_id, dtype: int64
id
291116    864691135373964873
262678    864691135323006620
292864    864691135497601299
298937    864691135587354052
292670    864691135303414951
294825    864691135815624783
292649    864691136965924814
260541    864691135777415357
301085    864691135941379188
264824    864691135100010784
303149    864691136031932603
Name: pt_root_id, dtype: int64

Note that these are not the same. Based on some other testing, old_roots is the correct answer.

[fcollman]

Yeah you need the supervoxel column, should figure out how to enforce that for live queries