[
https://issues.apache.org/jira/browse/ARROW-487?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15824338#comment-15824338
]
Wes McKinney commented on ARROW-487:
------------------------------------
Sounds right. I recommend acquiring the GIL in a base
{{PandasBlock::~PandasBlock}} and manually releasing / DECREF-ing the arrays
> Python: ConvertTableToPandas segfaults if ObjectBlock::Write fails
> ------------------------------------------------------------------
>
> Key: ARROW-487
> URL: https://issues.apache.org/jira/browse/ARROW-487
> Project: Apache Arrow
> Issue Type: Bug
> Components: Python
> Reporter: Uwe L. Korn
>
> My first guess: The segfault occurs in `OwnedRef::~OwnedRef` with a NumPy
> array. This might be either the cause of NumPy trying to release garbage
> objects (as the array is not filled yet) or because we don't have locked the
> GIL during the deconstruction.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
