CVS log for 2000/10/swap/llyn.py

Request diff between arbitrary revisions

merge in latest performance ideal

fix output bug

A month of changes

fix tabs

add fn:doc built-in, refactor log:content a bit; remove leftover diagnostic from xpath builtin

fixed BI_dtlit to match newXMLLiteral()
  (though I'd rather newLiteral() were fixed)

fixed sax2rdf to handle attributes without namespaces
in XML Literals

back to passing all tests

oops typo

untested. seems to work xml literal in each combination parse & gen in n3 and rdf.

@@@ BEWARE UNTESTED PARTIAL VERSION -- Introducing XML Literals as DOM objects

minor things to get tests to pass, so new for python2.5

unification rewrite part I

step 1 of changes to allow for backward chaining

proof that release process still works

fix c14n and filter bug

slight changes

fix semanticsImportWithClosure

some attempts at performance help

dont always canonicalize --- a hack

undo earlier fix; it was wrong

some more fixed

fix a couple bugs introduced

we seem to have (mostly) stabilized

Paw test now seems to work

fix for equality

fast test only checked

simple fix

fix owl:sameAs

starting to be able to prove log:supports

a zeroth iteration of log:supports

passes rein tests, and all stable tests!

more things now match

updates to query

more work in includes

work on includes and variable renaming

more work unifying

work in includes

Seems to work. Tested on plane DC->BOS with Yosi

boolean literals, and fixes for --why

grouping of statements by reason; Various bugs out, duplicate checking always in runLaborious. Passed regression.n3 and reason/details.tests

Remove -B kludge on log:semantics

ONLY FAST TESTS DONE.  check.py and why.py improved and unification improved.  Proof and check the judy example for PAW work

--n3=B flag introduced. --why improved but not perfect.

 a little cleanup

Revive --why proof generation and check.py. Recode unification a bit

Passes test/Makefile.  smartIn changed to interpretBuiltins. rdf:first rdf:rest special.

lots of strange things

improved builtins and inclusions

wow!

closer to a working query

fix a spelling error in the name of loadFormulaWithSubs[t]itution

put sparql2cwm where it belongs

add log:dtlit

rename decimal.py

add log:semanticsWithImportsClosure

term is now a new-style class

added dtlit built-in; see test in test/dt/dtlit1.n3
@@todo: connect to test harness

fix for strange = bug in sets

set builtins and log:filter . Tests will follow

make ImmutableSets immutable enough for me

coded around a nasty bug in python2.3 ImmutableSet

a first step in supporting n3 sets

the N3Set datatype now exists

merge in the performance branch. From here on out, cwm uses sets.

preformance --- change variables to sets

list ops

fix for reification

fixed weird nested list bug Tim found

I\'m ready to announce the n3 tests

Fix bugs in cwm --patch
diff.py now can generate weak deltas when necessary, when the graphs are not solid.
Offline working hack in webaccess.py
Fixed bug in lllyn.py with interning of datatyped literals giving diff objects for same type  sometimes.

fix for decimal. and the release Makefile

undo what I did last thing Friday. Those changes are nowhere near ready for the trunk

so I have stuff to work on at home

rdflib tests in the default test suite

as of right now, the tr people can use cwm again.

see if we can get referer tags

a profiler, and the results of running it on a particularly bad file

An attempt to get setup.py working
Those changes get very annoying and kludgy
Also, string:search is implemented (not tested yet)

Yosi

new buglist, and preliminary support for decimal
Note that, per discussion with danC, the default for numbers is still float

Yosi

Two things:

string:search, per danC's RFE
nodeID should now be fixed

Yosi

Test files that I have forgotten to add to cvs before now

fixed test system and release system

Merging automatic generation of filelists into the main branch

Note that the distribute step is not yet set up to use it!

Yosi

Reification continues

Reify, and dereify. Dereify still has problems

Yosi

Added working rdflib support, and a number of minor bugs fixed

Yosi

An attempt to have Cwm keep track of the files used in the regression tests
for you. It is not particularly beautiful

Yosi

--patch uses delta namespace
-- removed i18n tests
-- numerics canonicalized
see doc/changes.html

(See doc/changes.html)
on xml output, nodeID was incorrectly spelled.
update.py provides cwm's --patch option.
diff.py as independent progrem generates patch files for cwm --patch

See http://www.w3.org/2000/10/swap/doc/changes.html for details
- Regresssion test incorporates the RDF Core Positive Parser Tests except XMLLiteral & reification
- xml:base support was added in the parser.
- Use the --rdf=R flag to allow RDF to be parsed even when there is no enveloping <rdf:RDF> tag
- nodeid generated on RDF output
- Automatically generated terms with no URIs sort after anything which has a URI.
- Namespace prefix smarts on output - default ns used for that most frequently used.
- suppresses namespace prefix declarations which are not actually needed in the output.
- Cwm will also make up prefixes when it needs them for a namespace, and none of the input data uses one.-
- Will not use namespace names for URIs which do not have a "#". Including a "/" in the flags overrides.

Introduced webAccess.py with  urlopenForRDF which (for now) adds an accept header ofr application/rdf+xml.
This allows cwm to get RDF versions of the W3C web site mail archive search outputs.

Passes tests. notation3.py has utf-8 encoding on output. Non-ascii text is output as such in qnames and strings.

- more ustring fixes
- promoted comment that internURI is obsolete to a python warning
- fixed rawUri (@@needs test)

Passes regression test but regression test still ocntains /devel/WWW.  Minor fix to ToXML.py o in encoding area

Added @keyword support.
(eventually got python+expat to wrok on fink, with patch)
Trig functions are in, thanks to Karl, with some changes, but NOT in regeression.n3
see test/math/test-trigo.n3 for now.

Added Trigonometry Module

Merged runSmart branh, which now passes tests.
The main change is that query.n3 mapps the rules before executing them,
and then exeutes them in an order which takes account of the
interdependencies between rules.  Cyclic subsystems of rules are
isoplated, and then the rules and those subsystems formed into a DAG
and executed in a topological ordering.  This is not a full rete
system yet.
The regression test file infor and program retest.py have changed
a little, working toward resuse of tests between applications.

Passes regression test. (make release)

panic just deleted *.py, saved from editor... sigh

show dan - doesn't work

Passes regression test. (make release)

Passes regression test. (make release)

Passes regression test. (make release)

Bug fixes (using --think=; adding new data to kb after --think, etc) . passes regression tests.

Passes regression test. (make release)

List processing improves - see doc/changes.html

Passes regression test. (make release)

fixed bug with genid collisions. @@ I owe a test.

Re-engineered in places - See doc changes.html

Passes regression test. (make release)

Fix --ugly to just dump data but  do dump the forSomes

Changed llyn mostly, so that the forAll and forSome pseudo-statements
are no longer stored in the formulae. Instead, each formula has a simple
list of universal vars and a list of existentials.
This means that the query engin can no longer query for forAll x. Thisis no great loss.
Also, started separating the inference engine from the store, making it
use the stores (or arther Formul's in mostcases) API instead of knowledge of internals.
Not finished, but an improvement.
Aiming to separate the pretty printer too soon.

Added caommand line flaf --closure=flags  to make a run crawl the web for more
info as it finds references to new documents.
s p o flags allow it to crawl by looking up URIs in those positions.
i should do imports and r rules but testing is minimal.
test case is test/ferrell/myDingo.n3
enjoy

Added but not completely tests --closure

not enough args to evalObj caused failure of test
 49)  excercise error case of log:semanticsOrError

Minor fixes. log:forSome and log;forAll no longer generated in xml rdf o/p

purge forAll along with implies; register maths

turn types.StringTypes into tuple(types.StringTypes) so it works with
more versions of python.   (eg 2.2b2)

Numeric typres returned by math builtins. See http://www.w3.org/2003/04/08-cwm.html

Datatypes now used in math and time - see maths and times for old functionality - ordering changed on ntriples output - unicode test now works - some work on unicode in various places - see i18n/hiragana.n3

Datatypes added. N3 has special syntax for int and double, suppressed with --n3=n . RDF/XML input and N3 in and out done, but NOT XML output.

finished up Products.id=1003 to Products#id.1001 transition

switched from sql: to mysql:

catch XML wf errors in semanticsOrError

fix datatype of queried values

Passes regression test

Cascade (tims laptop) lid break pre-mend backup emergency commit

Passes regression test

Improved documentation for Formula in llyn.py

fix semanticsOrError call to evalObj; test in includes/semErr.n3

Passes 64 retests with -n -c

Integrate EricP's changes and my minor things

used query items in ResultSet and commentd remoteQuery

Passes python -n -f regression.n3

added test refs, prepared to obsolesce log:authoritativeService on sql: URIs

NOT TESTED

adjusted to new query interface, added log:pointsAt and demo query using it (test/dbork/OrderTracking1.n3)

SqlDB works on unknown DBs

Fix bug when query involves list with the same variable in  more than once

Works without proof stuff, working on proof stuff....

Simplied the interface to built-in function modules. retest ok

--why works up to reason/t5. GK and SBP's list bugs fixed.

cleanup only

some --flatten tests fail. --why fails. Formulae must be closed to be referenced in a add()

mmm

Rearranged proof ontology again...

retest ok, --why seems too work on simple cases

--why does NOT work

Passes test/retest.sh.  Minumum changes

base can be None in uripath:refTo

minor api changes probably

some minor changes to LX-formula stuff, but it's still broken in ways
that don't show up on the regression test.

mostly: moved llyn-LX conversion stuff out of llyn.py, into
LX.engine.llynInterface.py

Switched cwm main-loop to keeping state in llyn AND/OR an LX.formula,
as needed by each command-line option.  Also factored out common
language code in main loop, so cwm can handle more than just "rdf" and
"n3".  New functionality is not thoroughly tested, but old functionality
is and should be fine.  Also did a few changes to LX variable
handling.

make --flatten recognize log:means as the biconditional and
log:Falsehood for a formula as its negation.

added support for conversion to & from LX formulas

added SqlDB support to main branch
  retest and diff of swap/test/diffs/ indicated no change to tested code

SqlDB works with llyn

integrating SqlDB with cwm

Add two tests for  quick variable ?x syntax. Passes text/retest.sh.

Path parsing now tested too, simply.

RemoteQuery hooks ready in llyn.py. tests 100%

uripath changes. passes 51 general tests and 25 loopback tests

cleanup: diagnostic interface, relative/absolute uri handling