CVS log for perl/modules/W3C/Rdf/SqlDB.pm

Request diff between arbitrary revisions

+ addTriple takes an attribution
+ refreshNode internal API
+ RdfDB::clearAttribution (sets refreshNode)
~ fixed up AttrList code
~ stop double-inserting triples

~ _bindingsToStatements uses -assertionAttribution

~ fixed utf8 encoding issues

+ look out for W3C::Rdf::AlgaeCompileTree::Members types in unexpected places
+ place to write down ORDER BY
+ processDeleteion
+ special support for W3C::Rdf::AlgaeCompileTree::Members objects (only) in W3C::Rdf::SqlDB::HoldsProperty::processMatch
~ WildcardProperty::processMatch has proper cardinality when working only with __Holds__ table
~ update to current dbEvaluateAssertion API
+ dbEvaluateDeletion
~ re-use AttributionList joins with same variables names (on way to factoring it out)

+ {-assertionAttribution}

~ simplified Algae2::{addSource,getSourceByName} API

+ connected named graph (Ask) structure to attribution constraints
+ updated dbEvaluateAssertion to reflect current API

~ use correct data: synta
~ eliminate nextElement call on empty result set

+ handle constants from query parses

~ handle constraint terms (as opposed to constraints attached to terms)

ignore unimplemented constraints for now

keep properties around

clarify message

added datatype and unacessible predicate support

added datatype support

changed constructor API for Algae2

modern RdfApp constructor

get disjunction binder working naively

finishReuse returns the re-used disjunctions

migrated to Binder architecture
ready to work on DisjunctionBinder

begin migration to Binder architecture

clean up comments on binding constraints

clean up binding constraints

ganged up earlier solution constraints into a large disjunction

early warning error message

die if no props file is given

can't take refs even if they eval to scalars

constant bindings (those just for select) don't constrain

added constraint disjunction

added support for URI to field maps

added (constraint) support for already bound variables

fixed attribution result error (was returning as a scalar)

follow name change of Disjunction

add the $Id again

catch absorption errors and remove the screwed up row

separate generic triple store requirements from app-specific schema access

may not even have an auth

make now works in Rdf

enable loading of attributions when loading statements by key

provided api for primary keys

pass all the parms through

better error message for unknown field

probably made queries with an initial optional triple work

added support for floats

hacked delete to work, not done thinking

handle unselected attributions

fixed AttrList re-use and assign attribution constraints

fixed nth binding query underconstraints problem and added trivial attribution constraints

start on attributions tracking

re-orged prior to tackling attributions

more specific exception

s/GenId/BNode/g

use a constructor flag for overconstraints check

brute force no overconstraint checking, fix this later

follow the ResultSet iteration API

working with new Attribution structure

got rules working again, reworked Attributions for proofs

triplesMatching relies on graph machinery, handle NULLs from DB

getting annotea working with new attributions

made properties parameter more flexible

s/AlgaeStructure/AlgaeCompileTree/

switched to RdfDB for proofs, brought FingerDB up to date

switched to using the sanctioned absorb API

fixed last known Attributions->AttrList bug

working on AttribList

switched to AttributionList and cleaned out bagIDs

oops, clumsily forgot to test if it even needed a datatype

interesting corner case in __Holds__ will also rear fugly head in queries with subjects constrained by foreign keys

re-org'd a bit -- cleaning up

created DisjunctionBindings

toString defaults to making no assumptions about operator precedence

some comments on the commonClause code

switching to more rigourous object boundries

restructured wheres to reduce scalar pointers

serialize query according to operator precedence

cleanup

added partial support for wildcard predicates

fixed table name

disjunction works

zillion-fold improvement in performance

switched from hash-subset with collision count to simple foreign key for __Nodes__

algae2 now worky

changed default attributing variable name

tweaked grammer, API and algae query lang to handle datatypes

AlgaeStructure.pm

re-phrased throw and put limit control on ObjectDB

switched to more brief exception idiom

forgot to dereference column number to get value

making use of resolveUriRdfId

keep fabricated query piece from bollixing up result array

rearranged predicate-dependent handling of query construction

working on hetro DB

some comments on the upcoming code

moving hetero join implications to regular predicate handler

some code factoring and notes

first pass at heterogeneous db

fixed NULL bindings for SqlDB and renamed NULL

added aggregation of existential variables

isomorphism

extraneous variables

trying out objecty database descriptions

some notes

made LEFT OUTER JOINs work

changed some var names

re-orged for left outers

fixed lucky return value and brought some mysteriously absent notes back from seldom-seldom land

extraneous variable

extraneous variable

nuked vestigial variable

implemented overconstraints tests

ACL queries working

composed constant key identifiers work

reformatting and commenting

removed negated statements from resulting statements

constraint checking

re-organized walkQuery call to give disjunctions more intimate control

moved member variable initialization into constructor

experimenting with disjuction labels

eliminated spurious parens

better with symmetry to SCALAR_BINDINGS and OBJECT_BINDINGS? perhaps if later subclassed.

disjunction now works

got NOT working

moved wheres to a SqlDBAlgae member

moved many stack variables to SqlDBAlgae members

worked on whitespace and grouping for readability

worky

todo notes

DB access a bit underconstrained

fixed column number references and culled a bit

new column binding algorithm

found annoying print statement

deprecated varInfos

restructured and commented for readability

switched to references

moved class variable WHERES to stack

spurious variable elimination

made interface rules work

starting algae 'interface' directive implementation

added some cute formatting

now does assertions

added supporting statements

processQueries worky

reformatting for inheritence

improve error reporting

comments describing incremental group compilation

new RdfDB interface to SQL tabels