* globals.c libxml.h parser.c parserInternals.c tree.c xmllint.c
xmlreader.c include/libxml/parser.h: a lot of performance work
especially the speed of streaming through the reader and push
interface. Some thread related optimizations. Nearly doubled the
speed of parsing through the reader.
Daniel
* DOCBparser.c HTMLparser.c c14n.c catalog.c encoding.c globals.c
nanohttp.c parser.c parserInternals.c relaxng.c tree.c uri.c
xmlmemory.c xmlreader.c xmlregexp.c xpath.c xpointer.c
include/libxml/globals.h include/libxml/xmlmemory.h: added
xmlMallocAtomic() to be used when allocating blocks which
do not contains pointers, add xmlGcMemSetup() and xmlGcMemGet()
to allow registering the full set of functions needed by
a garbage collecting allocator like libgc, ref #109944
Daniel
* configure.in: switched to have thread support enabled by default,
didn't got troubles with ABI compatibility on Linux, hope it
won't break on strange OSes, if yes, report the system ID
* doc/libxml2-api.xml: just rebuilt the API
Daniel
* libxml.h include/libxml/parser.h parser.c xmlIO.c DOCBparser.c:
added support for large file, tested with a 3+GB instance,
and some cleanup.
* catalog.c: added a TODO
* Makefile.am: added some "make tests" comments
Daniel
* relaxng.c: some cleanups
* doc/xmlreader.html: extended the document to cover RelaxNG and
tree operations
* python/tests/Makefile.am python/tests/reader[46].py: added some
xmlReader example/regression tests
* result/relaxng/tutor*.err: updated the output of a number of tests
Daniel
* relaxng.c xmllint.c: more work on RelaxNG streaming validation
trying to improve the subset compiled, and more testing.
* doc/downloads.html doc/xml.html doc/xmlmem.html: some updates on the
documentation
* test/relaxng/tutor11_1_3.xml: fixes the DTD path
* result/relaxng/*.err: fix some of the outputs
Daniel
* relaxng.c xmlreader.c xmllint.c include/libxml/relaxng.h
include/libxml/xmlreader.h: implemented streaming of
RelaxNG (when possible) on top of the xmlReader interface,
provided it as xmllint --stream --relaxng .rng .xml
This seems to mostly work.
* Makefile.am: updated to test RelaxNG streaming
Daniel
* relaxng.c include/libxml/relaxng.h: integrated the regexp
based validity checking of fragments of the document for
which the RNG can be compiled to regexps. Works on all regression
tests, only fix needed is related to error messages.
Daniel
* relaxng.c xmlregexp.c include/libxml/xmlautomata.h
include/libxml/xmlregexp.h: Starting work precompiling
parts of RelaxNG schemas. Not plugged onto validity checking
yet, just the regexp building part. Needed to extend some
of the automata and regexp APIs.
Daniel
* xmlreader.c include/libxml/xmlreader.h: Added the Expand()
and Next() operation to work on subtrees within the reader
framework.
* doc/libxml2-api.xml python/libxml2class.txt: resulting updates
* python/tests/reader5.py: added an example for those new
functions of the reader.
Daniel
* xmlschemas.c xmlschemastypes.c include/libxml/xmlschemas.h:
update from Charles Bozeman for date and duration types
* test/schemas/date_0.* test/schemas/dur_0.*
result/schemas/date_0.* result/schemas/dur_0.*: updated too
Daniel
* tree.c valid.c xpath.c include/libxml/tree.h include/libxml/valid.h:
fixing bug #107129, removing excessive allocation and calls
to *printf in the code to build QName strings.
Daniel
* relaxng.c: try to work on bug #109225 and provide better
error reports.
* result/relaxng/* : this change the output of a number of tests
* xinclude.c: fixing the parsed entity redefinition problem
raised on the list.
* test/schemas/date_0.xsd: updated the date test c.f. E2-12
Daniel
* xmlschemastypes.c: fixed date comparison to handle the tzo
The only failures left are disagreements on Notations and
'+1' not being allowed for ulong, uint, ushort and ubyte.
Daniel
* relaxng.c xmlschemastypes.c: more work on XML Schemas datatypes
and facets support. Currently only schemas with binHex or
base64 don't compile. A few error left in the test suite:
found 1035 test instances: 919 success 23 failures
most are gdate or gdateyear failing check, and a few cases where
James clark tests results are strange.
* valid.c: allow to reuse the Notation checking routine without
having a validation context.
* SAX.c: removed a #if 0
Daniel
* relaxng.c include/libxml/relaxng.h xmlschemastypes.c: more work
on cleaning up XML Schemas datatypes based on James Clark tests
test/xsdtest/xsdtest.xml
Daniel
* relaxng.c: implemented comparisons for Schemas values.
* xmlschemastypes.c include/libxml/xmlschemastypes.h: fixed
some bugs in duration handling, comparisons for durations
and decimals, removed all memory leaks pointed out by James
testsuite. Current status is now
found 238 test schemas: 197 success 41 failures
found 1035 test instances: 803 success 130 failures
Daniel
* xmlschemas.c include/libxml/xmlschemas.h: fixed bugs and memory
leaks in the W3C XML Schemas code
* xmlschemastypes.c: implemented nonPositiveInteger
* test/schemas/length2_0.xsd result/schemas/length2_0_0.err:
fixed the test and result.
Daniel
* HTMLparser.c tree.c: two patches from James Bursa on the HTML
parser and a typo
* xmlschemastypes.c: reindenting, fixing a memory access
problem with dates.
Daniel
* parser.c: fixing #109227 providing more context in case of
start/end tag mismatch
* python/tests/ctxterror.py python/tests/readererr.py: update the
tests accordingly
Daniel
* xpath.c include/libxml/xpath.h: first part of the fix to
performance bug #108905, adds xmlXPathOrderDocElems() providing
document order for nodes.
* python/libxml.c: Python may require TRIO as Albert Chin pointed out
Daniel
* parser.c: fixing bug #108976 get the ID/REFs to reference
the ID in the document content and not in the entity copy
* SAX.c include/libxml/parser.h: more checking of the ID/REF
stuff, better solution for #107208
* xmlregexp.c: removed a direct printf, dohhh
* xmlreader.c: fixed a bug on streaming validation of empty
elements in entities
* result/VC/ElementValid8 test/VCM/v20.xml result/valid/xhtml1.xhtml:
cleanup of the validation tests
* test/valid/id* test/valid/dtds/destfoo.ent result/valid/id*:
added more ID/IDREF tests to the suite
Daniel
* SAX.c: Fixed validation bug #108858 on namespace names using
entities and reported by Brent Hendricks
* xmllint.c: report xmlTextReaderHasValue() result in --stream
--debug output.
Daniel
* xmlreader.c: fixed bug #108801 reported by Malcolm Tredinnick
about the DocType node not being reported sometimes.
* python/tests/reader.py: added to test to the regression checks
Daniel
* HTMLparser.c parser.c parserInternals.c: patch from
johan@evenhuis.nl for #107937 fixing some line counting
problems, and some other cleanups.
* result/HTML/: this result in some line number changes
Daniel
* configure.in Makefile.am: fixed Red Hat bug #86118 use libxml2.spec
instead of libxml.spec
* relaxng.c: fixed some of the error reporting excessive
verbosity
* catalog.c debugXML.c valid.c xmlreader.c xmlschemas.c xpath.c
xmlschemastypes.c: removed some warnings from gcc
* doc/libxml2-api.xml: rebuilt
Daniel
* relaxng.c: fixed xmlRelaxNGNodeMatchesList
* test/relaxng/testsuite.xml: augmented the test suite
* result/relaxng/spec1* result/relaxng/tutor12_1*: this fixes
some schemas validation tests in the presence of foreign
namespaces.
Daniel
* valid.c xmlschemastypes.c: attempt to cope with ID/IDREF(S)
declared both in the DTD and in the Schemas <grin/>
* relaxng.c: more debug, added a big optimization for <mixed>
* test/relaxng/testsuite.xml: augmented the testsuite
* test/relaxng/ result/relaxng: added the RelaxNG spec and a
DocBook example to the regression tests
Daniel
* tree.c: removed a warning
* xmlschemastypes.c: more cleanup, added ENTITY and ENTITIES
support
* check-relaxng-test-suite.py check-xsddata-test-suite.py:
cleanup/improvements of the regression tests batch
* test/relaxng/testsuite.xml: augmented libxml2 own testsuite
Daniel
* relaxng.c: fixed error msg cleanup deallocation
* xmlschemastypes.c: added a function to handle lists of
atomic types, added support for IDREFS
Daniel
* relaxng.c valid.c xmlschemastypes.c: added Datatype ID
and IDREF, usable from RelaxNG now
* include/libxml/xmlschemastypes.h: need to add a new interface
because the validation modifies the infoset
* test/relaxng/testsuite.xml: extended the testsuite
Daniel
* relaxng.c: fixed the last core RelaxNG bug known #107083,
shemas datatype ID/IDREF support still missing though.
* xmlreader.c: fix a crashing bug with prefix raised by
Merijn Broeren
* test/relaxng/testsuite.xml: augmented the testsuite with
complex inheritance tests
Daniel
* relaxng.c: switched back to the previous Relax-NG code base,
the derivation algorithm need severe constraining code to avoid
combinatorial explosion. Fixed the problem with Sebastian Rahtz
TEI based example and other bugs
* result/relaxng/*err: updated the results
* test/relaxng/testsuite.xml: started a new test suite
Daniel
* relaxng.c include/libxml/relaxng.h: After coming to the conclusion
that the original RelaxNG validation code was un-fixeable, it got
rewritten to use the derivation algorithm from James Clark and
redebugged it (nearly) from scratch:
found 373 test schemas: 372 success 1 failures
found 529 test instances: 529 success 0 failures
<sigh/>
Daniel
* SAX.c parser.c: fix some recursion problems introduced in the
last release.
* relaxng.c: more debugging of the RNG validation engine, still
problems though.
Daniel
* Makefile.am: stop generating wrong result file with * in name
* relaxng.c: fixing the include bug raised by Sebastian Rahtz
* result/relaxng/demo* test/relaxng/demo: added the tests from
Sebastian reproducing the problem.
Daniel
* test/xsdtest/xsdtest.xml uri.c: after and exchange with James
Clark it appeared I had bug in URI parsing code ...
* relaxng.c include/libxml/relaxng.h: completely revamped error
reporting to not loose message from optional parts.
* xmllint.c: added timing for RNG validation steps
* result/relaxng/*: updated the result, all error messages changed
Daniel
* xmlschemastype.c: made powten array static it should not be exported
* HTMLparser.c: fix bug #107361 by reusing the code from the XML
parser function.
* testHTML.c: get rid of valgrind messages on the HTML SAX tests
Daniel
* tree.c: fixed a node dump crash on attributes
* test/xsdtest/xsdtest.xml test/xsdtest/xsdtest.xsl: fixed
an URI test bug and get better output.
Daniel
* relaxng.c xmlschemas.c xmlschemastypes.c
include/libxml/xmlschemastypes.h: added param support for relaxng
type checking, started to increment the pool of simple types
registered, still much work to be done on simple types and
facets checkings.
Daniel
* relaxng.c: Fixed a couple of problem raised by James Clark
in bug #107083, the support for ID/IDREF/IDREFS at the WXS
datatype level still not fixed though.
Daniel
* tree.c: fixed xmlSetProp and al. when the node passed is not an
element.
* relaxng.c: fixed bugs 7.3 (though not complete) and memory leaks
found 373 test schemas: 369 success 4 failures
found 529 test instances: 525 success 4 failures
* check-relaxng-test-suite.py: added memory debug reporting
Daniel
* uri.c parser.c: some warning removal on Igor's patch
* tree.c: seems I messed up with #106788 fix
* python/libxml.c: fixed some base problems when Python provides
the resolver.
* relaxng.c: fixed the interleave algorithm
found 373 test schemas: 364 success 9 failures
found 529 test instances: 525 success 4 failures
the resulting failures are bug in the algorithm from 7.3 and
lack of support for params
Daniel
* parser.c: another fix for nodeinfo in entities problem
* tree.c entities.c: fixed bug #106788 from James Clark
some spaces need to be serialized as character references.
Daniel
* parser.c: fixing some nodeinfo in entities problem raised
by Glenn W. Bach
* relaxng.c: implemented the first section 7.3 check
* result/relaxng/*: updated the results
Daniel
* relaxng.c: fixed some problems in the previous commit
and finished implementing 4.16 rules checking
found 373 test schemas: 353 success 20 failures
found 529 test instances: 519 success 6 failures
* result/relaxng/*: updated the results
Daniel
* relaxng.c: implemented the checks from section 7.1, fixed
some of the 4.20 and 4.21 problems.
found 373 test schemas: 338 success 35 failures
found 529 test instances: 519 success 6 failures
* result/relaxng/*: updated the results
Daniel
* Makefile.am configure.in: patched to have shared libraries
for Python regression tests and static binaries for gdb debug
in my development environment
* relaxng.c: more bugfixes
found 373 test schemas: 296 success 77 failures
found 529 test instances: 516 success 8 failures
* result/relaxng/*: updated the results
Daniel
* xmlschemastypes.c: float/double check bugfix
* tree.c include/libxml/tree.h: exported a function for NMTOKEN
validation
* xmlreader.c: add a TODO for Jody
* relaxng.c: bugfix bugfix bugfix
found 373 test schemas: 300 success 73 failures
found 529 test instances: 507 success 10 failures
* result/relaxng/*: updated the results
Daniel
* relaxng.c check-relaxng-test-suite.py: more work on the
RelaxNG implementation conformance testing.
found 373 test schemas: 284 success 89 failures
found 529 test instances: 448 success 47 failures
* result/relaxng/*: updated the results
Daniel
* relaxng.c: more bug-hunting
* testRelax.c include/libxml/relaxng.h: added --tree to dump the
intermediate rng tree
* python/generator.py: patch from Stephane Bidoul to fix the generator
on python < 2.2
Daniel
* check-relaxng-test-suite.py relaxng.c: more testing on the
Relax-NG front, cleaning up the regression tests failures
current state and I forgot support for "mixed":
found 373 test schemas: 280 success 93 failures
found 529 test instances: 401 success 68 failures
* tree.c include/libxml/tree.h xmlschemastypes.c: finished and
moved the Name, NCName and QName validation routine in tree.c
* uri.c: fixed handling of URI ending up with #, i.e. having
an empty fragment ID.
* result/relaxng/*: updated the results
Daniel
* check-xinclude-test-suite.py: improved the script accordingly
to the XInclude regression tests updates
* xpointer.c: Implemented XPointer element() Scheme W3C PR of 13
November 2002
* result/XPath/xptr/chapterschildseq result/XPath/xptr/vidchildseq
test/XPath/xptr/chapterschildseq test/XPath/xptr/vidchildseq:
augmented the Xpointer testsuite for the element() scheme
Daniel
* check-xinclude-test-suite.py xinclude.c: Work on the W3C/NIST
regression tests for XInclude, improved the script, improving
XInclude error reporting mechanism
Daniel
* test/relaxng/OASIS/spectest.xml: OASIS RelaxNG testsuite
* check-relaxng-test-suite.py: python script to run regression
against OASIS RelaxNG testsuite
* relaxng.c: some cleanup tweaks
* HTMLparser.c globals.c: cleanups in comments
* doc/libxml2-api.xml: updated the API
* result/relaxng/*: errors moved files, so large diffs but
no changes at the semantic level.
Daniel
* xpath.c: fixed doc comment problems
* python/generator.py python/libxml_wrap.h python/types.c: adding
RelaxNG wrappers
* python/tests/Makefile.am python/tests/relaxng.py: added a specific
test of those early Python RelaxNG bindings
Daniel
* libxml.spec.in: fixes a libtool problem on AMD 64bits builds
* relaxng.c: found the validation problem I had with interleave
when not covering all remaining siblings
* Makefile.am test.relaxng/* result/relaxng/*: augmented the
testsuite and check the RNG schemas against the RNG schemas
given in appendix A
Daniel
* configure.in xmllint.c: I f...ed up the default configuration
of schemas and --relaxng option display in xmllint, pointed by
Morus Walter.
* xlink.c: Sean Chittenden pointed a couple of errors in the XLink
detection module, fixes bug #105374.
Daniel
* HTMLparser.c tree.c xmlIO.c: comments cleanups
* Makefile.am: use xmllint for doing the RelaxNG tests
* configure.in: preparing 2.5.2 made schemas support default to
on instead of off
* relaxng.c: removed the verbosity
* xmllint.c: added --relaxng option
* python/generator.py python/libxml_wrap.h: prepared the integration
of the new RelaxNG module and schemas
* result/relaxng/*: less verbose output
Daniel
* HTMLparser.c include/libxml/HTMLparser.h: applied HTML
improvements from Nick Kew, allowing to do more checking
to HTML elements and attributes.
Daniel
* relaxng: more work on name classes, the "validate all" schemas
seems to work now.
* test/relaxng/* result/relaxng/*: augmented/updated the
regression tests
Daniel
* python/libxml.c: removed an unprotedted debug message Aleksi Suhonen
* parser.c: put a guard against infinite document depth, basically
trying to avoid another kind of DoS attack.
* relaxng.c: some code w.r.t. nameClasses
Daniel
* relaxng.c: more work on Relax-NG, implementing externalRef
* test/relaxng/* result/relaxng/*: augmented/updated the
regression tests
* Makefile.am: cleanup to Relaxtests target
Daniel
Mon Jan 27 07:35:29 MST 2003 John Fleck <jfleck@inkstain.net>
* doc/tutorial/customfo.xsl
* doc/tutorial/customhtml.xsl
adding stylesheet customizations used to generate fo
for pdf and html
* relaxng.c: more work on Relax-NG
* test/relaxng/* result/relaxng/*: augmented/updated the
regression tests
* xmlschemastypes.c: added a number of base type definition but not
the associated checks, those are still TODOs
Daniel
Sun Jan 26 17:02:29 MST 2003 John Fleck <jfleck@inkstain.net>
* doc/xmlcatalog.1
* doc/xmlcatalog_man.html
* doc/xmlcatalog_man.xml
belatedly fixing bug #93622 (adds rewriteURI type to
"--add" option in xmlcatalog man page
* README: updated the policy on private mail answers
* relaxng.c: more work on Relax-NG
* test/relaxng/* result/relaxng/*: augmented/updated the
regression tests
Daniel
* relaxng.c: more work on Relax-NG
* doc/*: regenerated the docs
* test/relaxng/* result/relaxng/*: updated and augmented the
Relax-NG regression tests and results
Daniel
* Makefile.am configure.in relaxng.c include/libxml/relaxng.h:
First commit of the new Relax-NG validation code, not generally
useful yet.
* test/relaxng/* result/relaxng/*: current state of the regression
tests
Daniel
* xmlreader.c doc/apibuild.py: applied a new patch from
Stéphane Bidoul for cleanups
* doc/libxml2-api.xml: rebuilt the API description with
new entry points
Daniel
* xmlreader.c python/drv_libxml2.py python/generator.py
python/libxml.c python/libxml.py python/libxml_wrap.h
python/types.c: patch from Stéphane Bidoul for better per
context error message APIs
* python/tests/ctxterror.py python/tests/readererr.py:
update of the tests
Daniel
* xmlreader.c include/libxml/xmlreader.h python/generator.py
python/libxml.c python/libxml.py win32/libxml2.def.src: applied
a patch from Stéphane Bidoul to allow per XMLtextReader error
and warning handling
* python/tests/Makefile.am python/tests/readererr.py: adding the
specific regression test
Daniel
* python/generator.py python/libxml.c python/libxml.py
python/libxml_wrap.h python/types.c: applied and fixed a patch
from Stephane Bibould to provide per parser error handlers at the
Python level.
* python/tests/Makefile.am python/tests/ctxterror.py: added a
regression test for it.
Daniel
* xmlreader.c: fixed the streaming property of the reader,
it was generating tree faster than consuming it. Pointed out
by Nate Myers
* tree.c: fixed a bug in xmlSaveFormatFileEnc if passed a NULL doc
Daniel
* python/libxml.py python/libxml.c python/libxml2-python-api.xml:
fixed bug #102181 by applying the suggested change and fixing
the generation/registration problem.
Daniel
* parser.c: one more IsEmptyElement crazyness, that time in
external parsed entities if substitution is asked.
* python/tests/reader3.py: added a specific test.
Daniel
* HTMLtree.c tree.c: fixes#102920 about namespace handling in
HTML output and section 16.2 "HTML Output Method" of XSLT-1.0
* README: fixed a link
Daniel
* configure.in NEWS: preparing 2.5.0 release
* SAX.c: only warn in pedantic mode about namespace name
brokeness
* globals.c: fix a doc generation problem
* uri.c: fix#101520
* doc/*: updated and rebuilt the doc for the release, includuding
stylesheet update
* python/Makefile.am: fix a filename bug
Daniel
* tree.c : applied patch from Lukas Schroeder for register callbacks
* valid.c: modified patch from Lukas Schroeder to test
register callbacks with --chkregister
Daniel
* xmlreader.c: seriously changed the way data are pushed to
the underlying parser, go by block of 512 bytes instead of
tryng to detect tag boundaries at that level. Changed the
way empty element are detected and tagged.
* python/tests/reader.py python/tests/reader2.py
python/tests/reader3.py: small changes mostly due to context
reporting being different and DTD node being reported. Some
errors previously undetected are now caught and fixed.
* doc/xmlreader.html: flagged last section as TODO
Daniel
* xmllint.c: a bit of cleanup
* xmlreader.c: small fix
* doc/xmlreader.html: more work on the XmlTextReader tutorial
* python/libxml.py: a few fixes pointed out by Hannu Krosing
Daniel
