search

datasalt / splout-db

A web-latency SQL spout for Hadoop.
50 stars 14 forks source link

Automatic indexing of partitioning field #23

Closed navshaikh closed 11 years ago

navshaikh commented 11 years ago

If I look up the usage of Splout Tablespace View Generator as follows:

hadoop jar splout-*-hadoop*.jar simple-generate

the option '-idx' says: -idx - The additional field to index (fields used to partitioning are always indexed)...

This seems to imply that the partition key will be used for indexing automatically. However, it does not index this field by default unless indicated with -idx option.

I also verified it by getting the list of indexes on the SQLite file after tablespace is generated, which comes back empty.

sqlite> PRAGMA index_list(myTable)

Conceptually it would seem that the intended behaviour is to automatically index the partition key without having to indicate it as one of the fields to index.

pereferrera commented 11 years ago

Hi, thanks for reporting.

Indeed the documentation is wrong. It is a legacy sentence not true anymore. I committed the fix and the documentation will be clearer in the next release -