the 1010 query language is a state-based non-SQL query language, designed by joel kaplan and modeled on the IBM installed user program APLDI-II.

1010 is state-based insofar as queries are implemented as sequences of function-calls.  each function in the query takes one or more tables as inputs and produces a new table as outut.  input and output tables are global objects in the extnded workspace.

1010 uses a sophisticated execution model in which computed values are memoized and the computation of selections, joins, column-definition, and aggregation is deferred until actually required for continued execution of the query.

the Shakti model of 1010 is a subset of the complete language, implementing just enough of the full functionality to demonstrate both the power of the Shakti edition of k, and the expressive capabilities of 1010.

the core functions included in the Shakti model (1010.k) are:

  base    - denote a base table T
  sel     - select records in T
  link    - left join
  asof    - asof join
  willbe  - define a new column in T
  tabu    - define aggregations in T

three additional functions are provided:

  load    - map files in a directory of subprocesses
  get     - return the current state of the query
  set     - deposite arbitrary state in the subprocesses of a query

the model (mac only) is freely availble for study, and is found here:

  http://www.nsl.com/shakti/1010.zip

unzip the contents, which will create ~/1010

  1010.k
  1010.txt
  g.k
  k
  m.k
  show.k
  weather.k
    
  tq/
  csv/

run a few sample queries:

  cd ~/1010
  ./k 1010.k

the model employs a k implementation of map-reduce.  this can be found in a separate script:

  cd ~/1010
  ./k m.k

a simple query is traced through map-reduce.  press enter to advance to the next step in the computation.