Package com.illumon.iris.db.v2.updateby.fill

Class BooleanChunkedFillByOperator

java.lang.Object
com.illumon.iris.db.v2.updateby.fill.BaseDynamicFillByOperator
com.illumon.iris.db.v2.updateby.fill.BooleanChunkedFillByOperator
All Implemented Interfaces:
ChunkedUpdateByOperator
public class BooleanChunkedFillByOperator
extends BaseDynamicFillByOperator
A SafeCloseable object to carry forward context for a singular column between iterations of chunk blocks.

  • Constructor Details

  • Method Details

    • addChunk

      public void addChunk​(@NotNull ChunkedUpdateByOperator.UpdateContext inCtx, @NotNull LongChunk<Attributes.OrderedKeyIndices> keyChunk, @NotNull Chunk<Attributes.Values> valuesChunk, long groupPosition)
      Add a chunk of values to this table. The update process is essentially this:

      Iterate over the chunk of values: - If the value was null, then track that this row-key is redirected and set the value of the redirection to either, the last valid (non-null) row, or the Index null key, if no valid rows have been found

      - If the value was NOT null then - Update the redirections for this row, to this key - If we passed a currently redirected index: - Update every redirection up to the current key with the last valid key - Any of these that changed become output modified rows. - Set the last valid key to this one, and add this key to the set of redirections to remove.

      Parameters:
      inCtx - the context object for this update.
      keyChunk - the chunk of keys being added.
      valuesChunk - the chunk of values being added
      groupPosition -

    • modifyChunk

      public void modifyChunk​(@NotNull ChunkedUpdateByOperator.UpdateContext inCtx, @NotNull LongChunk<Attributes.OrderedKeyIndices> prevKeyChunk, @NotNull LongChunk<Attributes.OrderedKeyIndices> keyChunk, @NotNull Chunk<Attributes.Values> preValuesChunk, @NotNull Chunk<Attributes.Values> postValuesChunk, long groupPosition)
      Modify a chunk of values. This behaves much like a ChunkedUpdateByOperator.removeChunk(UpdateContext, LongChunk, Chunk, long) of previous values followed by an ChunkedUpdateByOperator.addChunk(UpdateContext, LongChunk, Chunk, long) of the current values with a few shortcuts.

      There are 4 possible scenarios

      1: A row was previously null, and stayed null: We don't actually care. The redirection is already correct so there's nothing else to do

      2: A row was previously null and became non-null Update the BaseDynamicFillByOperator.redirection to mark this row as not redirected and add it to the set of rows to be removed at the end of the cycle. Then, like an add, propagate forward any redirections between the last visited one, to this one

      3: A row was previously non-null and stayed non-null Nothing else to do here but check to see if we stepped past the next redirection index and propagate forward any redirections between the last visited one and this one

      4: A row was previously non-null and became null First, locate what the new redirection must be by finding the last valid key before the row being modified. Update the BaseDynamicFillByOperator.redirection with that value then look to the right and update any rows that were redirected to this one onto the new redirection.

      Parameters:
      inCtx - the context object
      prevKeyChunk - the chunk of previous keys
      keyChunk - the chunk of current keys
      preValuesChunk - the chunk of previous values
      postValuesChunk - the chunk of current values
      groupPosition -