[ 
https://issues.apache.org/jira/browse/BEAM-5487?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Luke Cwik updated BEAM-5487:
----------------------------
    Description: 
The definition of next for a byte key is incorrectly set as old key + 1 when 
treating the key as the next smallest value, for example

0x01 is the next key for 0x00

 

The ByteKey compareTo operator treats two keys where one key is the proper 
prefix of the other as smaller hence 0x01 is greater then 0x00, but so is 
0x0000 thus next is skipping over 0x0000 as the next smallest key larger then 
0x00.

  was:
The definition of next for a byte key is incorrectly set as old key + 1 when 
treating the key as the next largest value, for example

0x01 is the next key for 0x00

 

The ByteKey compareTo operator treats two keys where one key is the proper 
prefix of the other as smaller hence 0x01 is greater then 0x00, but so is 
0x0000 thus next is skipping over 0x0000 as the next smallest key larger then 
0x00.


> ByteKeyRangeTracker restrictions do not cover the entire interval because of 
> incorrect next key
> -----------------------------------------------------------------------------------------------
>
>                 Key: BEAM-5487
>                 URL: https://issues.apache.org/jira/browse/BEAM-5487
>             Project: Beam
>          Issue Type: Bug
>          Components: sdk-java-core
>            Reporter: Luke Cwik
>            Assignee: Luke Cwik
>            Priority: Major
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> The definition of next for a byte key is incorrectly set as old key + 1 when 
> treating the key as the next smallest value, for example
> 0x01 is the next key for 0x00
>  
> The ByteKey compareTo operator treats two keys where one key is the proper 
> prefix of the other as smaller hence 0x01 is greater then 0x00, but so is 
> 0x0000 thus next is skipping over 0x0000 as the next smallest key larger then 
> 0x00.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to