NAME

Genezzo::Block::RDBlk_NN.pm - Row Directory Block Not Null (array) tied hash class.

This class converts the Genezzo::Block::RDBlkA operations from a conventional array to a "Not Null" array. Genezzo::Block::RDBArray uses this class as the basis of a tied array.

Note: Like its parent RDBlock, this class is almost, but not quite, a pushhash.

SYNOPSIS

 use Genezzo::Block::RDBlk_NN;
 use Genezzo::Block::Std;

 local $Genezzo::Block::Std::DEFBLOCKSIZE = 500;

 my $buff = "\0" x 500; # construct an empty byte buffer

 my %tied_hash = ();

 my $tie_val = 
    tie %tied_hash, 'Genezzo::Block::RDBlk_NN', (refbufstr => \$buff);

 # pushhash style 
 # (note that the "PUSH" pseudo key is not supported)...
 my $newkey = $tie_val->HPush("this is a test");

 # or array style, your choice
 my $pushcount = $tie_val->PUSH(qw(push lots of data));

 $tied_hash{$newkey} = "update this entry";

 # a hash that supports array style FETCHSIZE
 my $getcount = $tie_val->FETCHSIZE(); # Note: not HCount

 # splice it
 my $err_str;

 my @a1 = $tie_val->HeSplice(\$err_str, 5, 3, qw(more stuff to splice));

DESCRIPTION

While RDBlkA adds array-like splice capabilities to RDBlock, violating the standard hash abstraction, RDBlk_NN violates the array abstraction, creating an array subclass called a "Not Null" array, a sort of priority queue.

"Not Null" arrays

Not Null arrays are designed to store non-null entries. Entries in the array which correspond to deleted values or metadata (see Genezzo::Block::RDBlock) may sometimes get returned as null entries, so various array manipulation algorithms should be adjusted to reflect this quirk. RDBlkA performs a good approximation of a well-behaved array, but the requirement to map a strictly ascending series of array offsets onto the normal data in RDBlock, which may be interspersed with deleted values and metadata, performs at O(n), versus O(1) for the "Not Null" version.

DELETE: Delete can have two outcomes: a deleted entry can disappear (standard outcome), or it can leave a null entry as a "hole".

FUNCTIONS

RDBlk_NN support all standard hash operations, with the exception that you cannot create or insert a user key -- you must push new entries and use the generated key or basic iteration to retrieve your data.

In addition to the RDBlock standard public methods, RDBlk_NN adds HSplice and HeSplice.

HSplice this, offset, length, LIST

Perform the equivalent of "splice" on the array.

offset is optional and defaults to zero, negative values count back from the end of the array.

length is optional and defaults to rest of the array.

LIST may be empty.

Returns a list of the original length elements at offset.

HeSplice this, error_ref, offset, length, LIST

error_ref is a string ref. Normally set to undef, set to error string on failure.

EXPORT

none

TODO

build simple test cases
build complex test cases
test thoroughly
packdeleted: make this work. It's broken!
integration with bt2 - need to packdelete in bsplit, do null checks in leaf blocks (branch blocks should be ok)
need a validation function to ensure that block maintains invariant: small number of leading metadata rows starting at row zero, followed by data rows (deletes ok). Easier to support non-split rows initially, but should be able to support head rows (need mods to splice functions to preserve rowstats for this case).
need to modify metadata methods so all metadata created in first n rows.
could simply have delete really delete the rows, so no changes necessary for rdblock clients (i.e., no "null rows" generated).

AUTHOR

Jeffrey I. Cohen, jcohen@genezzo.com