search

MLnick / hive-udf

Approximate cardinality estimation with HyperLogLog, as a Hive function
Apache License 2.0
42 stars 16 forks source link

How to make the UDAF run with old version of Hive ? #4

Closed moizarafat closed 4 years ago

moizarafat commented 11 years ago

Hi,

Had a question

a) What changes will be required to run this UDAF with older version of hive say 0.7

Thanks

MLnick commented 11 years ago

Hi Unfortunately I don't know offhand as I no longer really use Hive day-to-day. Have you tried it with 0.7? Does it throw errors?

moizarafat commented 11 years ago

Hi,

Some one already had that issue raised i think. "java.lang.NoSuchFieldError: writableBinaryObjectInspector while running the udaf " . I think its because there is no Binary Data type in hive 0.7. I was wondering if there is a workaround available.

MLnick commented 11 years ago

Ah

It would be possible to serialize the binary blob to say Base64 string. That would require code changes to the UDAF appropriately.

Any chance you can upgrade to newer Hive version? :)

— Sent from Mailbox for iPhone

On Thu, Sep 19, 2013 at 7:39 AM, maxwell1985 notifications@github.com wrote:

Hi,

Some one already had that issue raised i think. "java.lang.NoSuchFieldError: writableBinaryObjectInspector while running the udaf " . I think its because there is no Binary Data type in hive 0.7. I was wondering if there is a workaround available.

Reply to this email directly or view it on GitHub: https://github.com/MLnick/hive-udf/issues/4#issuecomment-24718202

moizarafat commented 11 years ago

Ok . Our prod environment is Hive 0.7 and we are still 4-5 months away from upgrade. We used the UDAF in our Test environment and worked great. we wanted to start using it in prod, but got stuck with 0.7 issue :(