[ https://issues.apache.org/jira/browse/ARROW-2288?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16391052#comment-16391052 ]
Antoine Pitrou commented on ARROW-2288: --------------------------------------- As for the crash: since {{Array::Slice}} adjusts the length when too large, it would make sense for it to also adjust the offset instead of crashing, IMO. > [Python] slicing logic defective > -------------------------------- > > Key: ARROW-2288 > URL: https://issues.apache.org/jira/browse/ARROW-2288 > Project: Apache Arrow > Issue Type: Bug > Components: Python > Affects Versions: 0.8.0 > Reporter: Antoine Pitrou > Assignee: Antoine Pitrou > Priority: Major > > The slicing logic tends to go too far when normalizing large negative bounds, > which leads to results not in line with Python's slicing semantics: > {code} > >>> arr = pa.array([1,2,3,4]) > >>> arr[-99:100] > <pyarrow.lib.Int64Array object at 0x7f550813a318> > [ > 2, > 3, > 4 > ] > >>> arr.to_pylist()[-99:100] > [1, 2, 3, 4] > >>> > >>> > >>> arr[-6:-5] > <pyarrow.lib.Int64Array object at 0x7f54cd76a908> > [ > 3 > ] > >>> arr.to_pylist()[-6:-5] > [] > {code} > Also note this crash: > {code} > >>> arr[10:13] > /home/antoine/arrow/cpp/src/arrow/array.cc:105 Check failed: (offset) <= > (data.length) > Abandon (core dumped) > {code} -- This message was sent by Atlassian JIRA (v7.6.3#76005)