Incorrect Values Rendered with uint8 Column on Datagrid #1084
Closed
Description
Bug Report
Steps to Reproduce:
- Create an arrow file that includes a uint8 column
- Pass the
ArrayBufferdata toperspectiveViewer.update(...)
Expected Result:
Rendered datagrid should be equivalent to what is printed with its pd.Dataframe and pa.Table equivalents. The first two sections show the pd.Dataframe values, pa.Table slots, and pa.Table.schema:
<Dataframe.info(memory_usage='deep')>
col_0 col_1 col_2 col_3 col_4 col_5
0 2020-05-19 name 0 0 0 0
1 2020-05-19 name 0 0 0 0
2 2020-05-19 name 0 0 0 0
3 2020-05-19 name 0 0 0 0
4 2020-05-19 name 0 0 0 0
5 2020-05-19 name 0 0 0 0
6 2020-05-19 name 0 0 0 0
7 2020-05-19 name 0 0 0 0
8 2020-05-19 name 0 0 0 0
9 2020-05-19 name 0 0 0 0
10 2020-05-19 name 0 0 0 0
11 2020-05-19 name 0 0 0 0
12 2020-05-19 name 0 0 0 0
13 2020-05-19 name 0 0 0 0
14 2020-05-19 name 0 13966 -13966 1
15 2020-05-19 name 0 0 0 0
16 2020-05-19 name 0 0 0 0
17 2020-05-19 name 0 1 -1 0
18 2020-05-19 name 0 0 0 0
19 2020-05-19 name 0 -39421 39421 1
--------------------------------------------------------------------
[<pyarrow.lib.ChunkedArray object at 0x7f7d540c8d68>
[[...]],
<pyarrow.lib.ChunkedArray object at 0x7f7d540c8368>
[[...]],
<pyarrow.lib.ChunkedArray object at 0x7f7d540c8818>
[[...]],
<pyarrow.lib.ChunkedArray object at 0x7f7d540c8c28>
[[...]],
<pyarrow.lib.ChunkedArray object at 0x7f7d540c8bd8>
[[...]],
<pyarrow.lib.ChunkedArray object at 0x7f7d540c89a8>
[[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,1]]
]
column_0: date32[day]
column_1: string
column_2: int32
column_3: int32
column_4: int32
column_5: uint8
Actual Result:
What the last section shows what is rendered on the browser-side. Notice col_5 has a difference in values in multiple rows (eg. row 4 and row 5), compared to above:
Environment:
Running Perspective webpack plugin, datagrid, and d3fc on the latest (0.5.0).
Additional Context
From briefly looking over the source code, could this be a casting issue (ie. in emscripten.cpp)?