Created 06-01-2016 03:43 AM
I am debating between using a column as string or date/timestamp. The field is a timestamp. however I see many jiras opened on hive date/timestamp which makes me believe I must ask the community prior to pushing forward. Of course all needs to be tested. However I want the communities feedback whether to avoid or not.
Created 06-01-2016 09:23 AM
There are many issues with timestamps, you're right, especially for older versions of Hive. Anyway, I'd suggest you to try to use timestamps. If you'll face any issue, you can ask the community or decide to revert to string...
Created 06-01-2016 06:50 PM
I have used time stamp as a string as well as date as well as timestamp, Advantages of using date and time is you could use date and time functions like unixtimestamp(), year() etc.. within hive this is much efficient than using string, especially if you have to do any sum bys or group bys.
Created 06-01-2016 07:01 PM
I have been using date and timestamp for some time and haven't run into any issues. In fact, for a customer we recommended moving away from string to timestamp for getting better compression.
If you can list all the JIRAs in your question, we can look at each one to see if you will run into issues on latest hive.