Thank you, Wes.

What about adding separate LZ4_BLOCK and LZ4_FRAME definitions in 
parquet-format, or at least tightening up the specification?

And is there any motivation in JVM parquet implementations to clear this up?

   michael


> On Oct 23, 2019, at 5:28 PM, Wes McKinney <[email protected]> wrote:
> 
> We are waiting on a volunteer to come forward and sort out the proper
> implementation in Parquet C++ for LZ4. It is disabled in the meantime
> I think
> 
> On Wed, Oct 23, 2019 at 10:11 AM Michael Heuer <[email protected]> wrote:
>> 
>> Hello,
>> 
>> There are a few issues related to block-vs-frame LZ4 compression in 
>> Parquet/Arrow and related
>> 
>> https://issues.apache.org/jira/browse/PARQUET-1241 
>> <https://issues.apache.org/jira/browse/PARQUET-1241>
>> https://issues.apache.org/jira/browse/PARQUET-1515 
>> <https://issues.apache.org/jira/browse/PARQUET-1515>
>> https://issues.apache.org/jira/browse/HADOOP-12990 
>> <https://issues.apache.org/jira/browse/HADOOP-12990>
>> https://github.com/dask/fastparquet/pull/315 
>> <https://github.com/dask/fastparquet/pull/315>
>> https://issues.apache.org/jira/browse/IMPALA-8617 
>> <https://issues.apache.org/jira/browse/IMPALA-8617>
>> 
>> Might anyone be able to summarize the current status and take a guess at the 
>> way forward?
>> 
>> Thank you in advance,
>> 
>>   michael

Reply via email to