Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

120 check-ins using file extensions/example/tcl.m4 or extensions/tnc/tcl.m4 or unix/tcl.m4 version 9404ac273b

2002-06-02
06:31
Added domlocks.c file check-in: 575b3d81c4 user: zoran tags: trunk
2002-05-26
19:55
Fixed the fix about var scopes - the parameters set with with-param must of course not be visible in the scope of the caller (call-template, apply-templates). Better handling of error cases: pop the var frame always befor checking the result code of ExecActions. This should guarantee var memory cleanup even in error cases therefor prevents (even more) memory leaks. Don't free the baseURI's subdoc entries for the xml and xslt document, because this strings are owned by the documents. check-in: adaa1ca752 user: rolf tags: trunk
19:54
Prevent the extResolver Tcl_Obj from shimmering by copying. (I wonder, why incrementing the ref Count is not enough, but I got "UpdateStringProc should not be invoked for type cmdName" panics while stress testing, which vanished, after this change.) check-in: a95866d02a user: rolf tags: trunk
19:52
Better handling of xslt -parameter option (copying of the parameter list guarantees, that the pointer to the string reps of the list obj's remain valid, even if somebody tries to cheat or does something unusual at tcl level in some handler script). check-in: 5a21f73136 user: rolf tags: trunk
2002-05-24
13:51
Closed some really big leaks (incomplete cleanup of subdocs). Plus better cleanup in case of xslt errors. check-in: fa50653391 user: rolf tags: trunk
13:48
Copy the extResolver for every dom document. The reason, why this is needed, is deep inside of tcl's handling of Tcl_Obj's. Inrecementing the obj's refCounter doesn't protect the obj to 'shimmer' (to change his internal representation). TclCmdNameObj's don't know how to regenerate there string representation, if that string rep was invalidated. If that happens, tcl panics with "UpdateStringProc should not be invoked for type cmdName". check-in: 45964023c8 user: rolf tags: trunk
2002-05-23
22:27
Better code for xslt -parameter handling check-in: 1a0f855ef3 user: rolf tags: trunk
22:25
Fixed bug with var scope (a var/param is only visible inside his defining template). Closed memory leak (don't strdup -parameter names). Better cleanup in case of xslt errors. check-in: 1a77be3578 user: rolf tags: trunk
22:19
Added deleting of the created parser objects, for lesser 'noise', while memory leak debugging. New test in xslt (var scope bug). check-in: b5dd1d0555 user: rolf tags: trunk
2002-05-22
16:42
Added deleting of the created parser objects, for lesser 'noise', while memory leak debugging. check-in: 50118713e6 user: rolf tags: trunk
16:40
Fixed a memory leak (handling of element content models). check-in: 6b68d7b876 user: rolf tags: trunk
2002-05-19
00:41
Optimization for predicate handling. No visible effect with standard xsltmark test files, but a lifesaver for other cases out of the wild. For example, xmlconf.xml / xmlconformance.xsl out of http://www.oasis-open.org/committees/xml-conformance/suite-v1se/xmlconf-20010315.htm is speeded up more than six times. check-in: adc606e802 user: rolf tags: trunk
2002-05-17
14:24
With space is significant, in source documents. check-in: 4299a6eab1 user: rolf tags: trunk
2002-05-16
20:36
External entities, needed by the dtd tests. check-in: f59c38b4f0 user: rolf tags: trunk
20:34
Added helper proc's, that read xml declaration, detect the encoding of the XML data and configures the input channel. check-in: 3e0659712a user: rolf tags: trunk
20:28
Removed reference to TclXML. Use [package require tdom] instead of [package require expat 2.0]. check-in: af8232dbf8 user: rolf tags: trunk
17:45
Added a xpath test non US-ASCII chars (uh, blame me, only one...) check-in: a25ac56080 user: rolf tags: trunk
17:30
Added tests for setting of global parameters. check-in: 33e13ca4d3 user: rolf tags: trunk
17:10
Updated tests to the changed createDocumentNS syntax. Added some tests. check-in: eed5e62bf7 user: rolf tags: trunk
13:16
There's something wrong, with the header files (well, at least VC++6.0 thinks so). Seems, it works in this include order. check-in: e60fa09baf user: rolf tags: trunk
12:46
Corrected version numbers. check-in: bda233a9c9 user: rolf tags: trunk
12:03
Corrected tdom stubs table export. check-in: d3a7fead57 user: rolf tags: trunk
2002-05-15
23:21
Updated after tdom.decls change (we don't have a hook). check-in: 36c50120a9 user: rolf tags: trunk
22:56
Changes needed for MS build. check-in: 9fde8342b2 user: rolf tags: trunk
22:54
Export tdom stubs table. check-in: cee75cca67 user: rolf tags: trunk
22:46
Inital commit check-in: c511c160a6 user: rolf tags: trunk
22:12
Use tdom stubs table. check-in: caa786e427 user: rolf tags: trunk
21:21
Initial check in. check-in: e787af70d9 user: rolf tags: trunk
20:19
Simplified QName check for creation of elements with xsl:element. Added QName check for creation of attributes with xsl:attribute. Report unresolvable prefixes in xml:strip-space / xml:preserve-space. check-in: 9c7ef4e572 user: rolf tags: trunk
20:15
Inital commit check-in: 35df4f33d8 user: rolf tags: trunk
19:42
New files needed for the tdom stubs table. (Under unix, tDOM still isn't build as stubs enabled extension, because that's a lot of define hassle, as log as Jochen need tcl8.0.5 support. But the extension stubs table is needed for the MS plattform, to make tDOM extensions - like tnc - work.) check-in: fcd3a5fdda user: rolf tags: trunk
16:19
*** empty log message *** check-in: 7913817752 user: rolf tags: trunk
2002-05-13
23:24
Ignore check-in: 2ebe7a8cbd user: rolf tags: trunk
2002-05-11
16:57
Made variable/parameter namespace aware. check-in: 53f9d10272 user: rolf tags: trunk
2002-05-10
21:52
Made modes namespace aware. check-in: 80c514621e user: rolf tags: trunk
20:29
Made key names namespace aware. check-in: 58cd284453 user: rolf tags: trunk
02:30
A few things at one: Made attribute set names namespace aware. If a literal result node has no namespace, and at the insertion point of the result tree is a default namespace in scope, unset the default namespace, while adding the node. Enhanced domSetDocument, that it not only set the ownerDocument right, but also resets the namespace indexes. check-in: 9e4ae2ec82 user: rolf tags: trunk
2002-05-09
01:17
Futher i18n-lization. Most basic stuff in xpathLexer() should be done no, though there is much room left for optimations. check-in: f376f8d74e user: rolf tags: trunk
2002-05-07
23:44
A NMTOKEN is always != the empty string (""); NMTOKENS are always at least one NMTOKEN. check-in: 559305970e user: rolf tags: trunk
2002-05-04
01:30
Inlined xpathRSInit (for speed). check-in: 378cc65ccb user: rolf tags: trunk
01:29
Some optimizations and a bit code cleanup. Removed all XPath functions code from xpathEvalStep() and packed it in a own xpathEvalFunction() function. Introduced a xpath function tag for fast identifying of the function; the tagging is done at xpath expression parse time (since the function name is already known, at this time) and therefor saves any strcmp() calls for any following calls to the function. Stripped the initialization of xpathResultSet's down to setting the type (to EmptyResult) and the nr_nodes (to 0) - that's all needed settings - and inlined the xpathRSInit(). check-in: 04590fa80c user: rolf tags: trunk
2002-05-03
01:23
Improvement for XPath substring(): Really only a corner case: postive infinity is a acceptable value for the third arg. (See XPath rec 4.2, it has an explizte example for this.) check-in: f5347c9ef2 user: rolf tags: trunk
2002-05-01
00:54
Introduced AxisDescendantLit and AxisDescendantOrSelfLit (a bit of a, aehm, let't call it workaround). With that, wie can distinguish between // and descencant or descendant-or-self. And reinsert the handling of the predicate filter with respect to the child axes for // in xpathEvalStepAndPredicates(). check-in: 24fb8a62ad user: rolf tags: trunk
2002-04-30
00:32
Better error detection in xsl:element: ":foo" isn't really a QName. Damned complicated check, for this single corner case, sigh. check-in: 883658ca7c user: rolf tags: trunk
00:26
Bug fix for XPath local-name(): the local name of a default namespace attribute (xmlns=".."") is the empty string "" (XPath rec 5.4) check-in: a6793350a4 user: rolf tags: trunk
2002-04-29
23:37
Bug fix for literal result elements and namespace copy from source to result: of course copy only the actual scope, not hidden namespaces (that ended in not wellformed results)! I thought, that is done already, but I simply was dreaming. Now we have this four interleaved while loops, which is quit a mess, in case of pathologic documents, like very deep one, lot's of namespaces declarations etc. check-in: 5c727a1687 user: rolf tags: trunk
20:49
Bug fix in domAppendNewElementNode(): if element uri is NULL, but there is a default namespace in scope, add xmlns="" (unset default namespace). check-in: c264ddc33a user: rolf tags: trunk
16:42
Bug fix in count(). Bug fix: corrected typo in unparsed-entity-uri() function name. check-in: 1de23925e3 user: rolf tags: trunk
16:26
Made xsl:call-namespace namespace aware. check-in: 3f260eddd1 user: rolf tags: trunk
15:56
Better propagation of (+/-)Infinity. check-in: bc58a13476 user: rolf tags: trunk
14:04
Cleanup in xpathEvalStep(): Use consistently xpathEvalStep() for every argument of XPath functions (instead of several xpathEvalSteps()). This is save, because XPath function calls are added with AddChildWithEvalSteps() in Production(FilterExpr). check-in: 003b2a37da user: rolf tags: trunk
03:29
Better propagation of NaN. check-in: c2d5fcb1eb user: rolf tags: trunk
02:29
Bug Fix: namespace attributes are not in a namespace. check-in: 92244eac07 user: rolf tags: trunk
02:12
Bug fix for XPath substring(). The type change of the var 'from' from unsigned int to int prevents an error, if from must be negative in the substring code. A bit code cleanup (and added handling of EmptyResult for two of the XPath functions. check-in: ba9ff2790f user: rolf tags: trunk
2002-04-28
23:56
Bug fixes in new UnaryMinus and full qualified attribute queries code. check-in: e1bad840e7 user: rolf tags: trunk
23:26
Bug fix for xpath round(): round(-2.5) should return -2. check-in: 9490c37af4 user: rolf tags: trunk
22:27
Improved xsl:elements: non QNAME name as element name is detected now. Bug Fix in domSetAttributeNS(). Small improvement of domCopyTo(): don't copy namespace attribute if it isn't necessary. Bug fix for xsl:copy. check-in: a9500c3a2b user: rolf tags: trunk
22:20
Added full qualified and namespace wildcard attribute queries. Fixed a Bug with following axis expression starting from an attribute. Added handling of UnaryExpr to xpathEvalStep(). Improved xpathGetPrio(). check-in: 48fafe5882 user: rolf tags: trunk
2002-04-26
01:14
Improved namespace support. New domCopyTo() for XSLT. Little improvement of xpathGetPrio(). check-in: 6925b06892 user: rolf tags: trunk
2002-04-23
23:55
Bug fix for following axis starting from an attribute. check-in: 40f2ded91f user: rolf tags: trunk
23:41
Bug Fix in xpathMatches check-in: 12a19a41ee user: rolf tags: trunk
22:01
Improved namespace handling for xsl:attribute. Improved attribute node set handling. Improved error detection: Unknown xslt elements and top level elements not in any namespace are reported. check-in: ae97545ffc user: rolf tags: trunk
20:46
domSetAttributeNS: if the attribute is in a namespace, check, if the namespace, actually bound to the attribute prefix is in deed the same, the attribute must be in. check-in: d7989bd111 user: rolf tags: trunk
20:42
Bugfix for FillNodeList: a match pattern with predicate may need a FillNodeList. Such a match pattern can now match the documentElement. check-in: 5a2a55be3e user: rolf tags: trunk
2002-04-22
21:33
Editorial changes. check-in: ab569aef00 user: rolf tags: trunk
21:32
Added xsl:namespace-alias. check-in: 5b62c9f047 user: rolf tags: trunk
00:54
Improved handling of literal result elements: now namespaces in scope are also copied to the result tree, if needed. exclude-result-prefixes and extension-element-prefixes of xsl:stylesheet elements are respected. (Still to do: xsl:extension-element-prefixes and xsl:exclude-result-prefixes attributes of literal elements.) check-in: 4ff4ec642f user: rolf tags: trunk
2002-04-19
20:48
Added handling of IS_NS_NODE flag set for an attribute node at a few more places. check-in: 084a5878bb user: rolf tags: trunk
18:55
Changed / enhanced namespace handling and namespace information storage. The namespace field of the domNode and domAttributeNode structurs is still set. But other than up to now, namespace attributes are now stored in the DOM tree as other, 'normal' attributes also, only with the nodeFlag set to "IS_NS_NODE". It is taken care, that every 'namespace attribute' is stored befor any 'normal' attribute node, in the list of the attributes of an element. The still saved namespace index in the namespace field is used for fast access to the namespace information. To speed up the look up of the namespace info, an element or attributes contains to, the namespace index is now the index number (plus offset 1) of the corresponding namespace info in the domDoc->namespaces array. All xpath expressions with the exception of the namespace axes (still not implemented) have to ignore this 'namespace attributes'. With this enhanced storage of namespace declarations, it is now possible, to find all "namespaces in scope" of an element by going up the ancestor-or-self axis and inspecting all namespace declarations. (That may be a bit expensive, for documents with lot of namespace declarations all over the place or deep documents. Something like http://linux.rice.edu/~rahul/hbaker/ShallowBinding.html (thanks to Joe English for that url) describes, may be an idea, if this new mechanism should not scale good enough.)

Changes at script level: special attributes used for declaring XML namespaces are now exposed and can be manipulated just like any other attribute. (That is now according to the DOM2 rec.) It isn't guaranteed (as it was), that the necessary namespace declarations are created during serializing. (That's also DOM2 compliant, if I read it right, even if this seems to be a bit a messy idea.) Because the old behavior have some advantages, from the viepoint of a programmer, it eventually should restored (as default or as 'asXML' option?). check-in: b5afde4623 user: rolf tags: trunk

2002-04-10
02:48
Optimization: Now handles xsltVarFrames and xsltVariables as stacks, instead of constantly malloc/free it. On the way removed xsltVariable->select (because it isn't used anywhere) and renamed xsltVariable->value to xsltVariable->node (because that's, to what this structur member points). check-in: e0fda7769a user: rolf tags: trunk
2002-04-09
17:20
Actual HEAD does not compile (!! arg). With this (intermediate) state it should. Some of the changes will be undo and replaced by others. check-in: e3ba6d222a user: rolf tags: trunk
01:59
xslt rec 5.2: "The string-value of an element node is the concatenation of the string-values of all text node descendants of the element node in document order." (Also true for root node.) The xpathGetTextValue() could not used for recursive getting of the text node descendants, because it also returns the text value of comment and processing instruction nodes. xpathGetTextValueForElement() does, what the spec requires. check-in: 69ed879f48 user: rolf tags: trunk
01:17
Corrected the check in log for the least check in... check-in: 78779edd38 user: rolf tags: trunk
01:14
Allowed reseting of variable in successive for-each loops (of course not in one loop) and inside template bodys of xsl elements. xt, xalan (and gnomexslt, for what is's worth) do, saxon don't - and the spec isn't really clear about this (I would say, it's right now). Anyway, there are only two clear ways, to handle this: allow it, or don't. tDOM does, but variable declarations inside the for-each (or other template bodies of xslt elements) polluted the var context outside the for-each or template bodies, which is clearly wrong (example: xalan tests: variable20, local: varscope). Unfortunately, the XSLT 2.0 WD is a bit clearer about this and does it the saxon way (no wonder, since Kay is the editor of that spec ;-)). But then XSLT 2.0 is another story, where currently definitely not in. check-in: 691495afd1 user: rolf tags: trunk
2002-04-08
03:43
Optimation: re-initialize current var frame in ApplyTemplates() loop only if necessary (i.e. if the applied template has "populated" it). check-in: 30ab7af562 user: rolf tags: trunk
02:11
Added -parameters option to domNode xslt method, to enable setting of top level parameters from tcl level. check-in: cf4305e95b user: rolf tags: trunk
02:01
Added optional -parameters option to domNode xslt method, to enable setting of top level parameters from tcl level. BugFix: Don't use direct calles to ApplyTemplate(), always go throw ApplyTemplates(), to get var context right (example: xalan tests variable35). check-in: 123891f1e6 user: rolf tags: trunk
2002-04-06
18:05
Bug fix for attribute value templates: "A right curly brace inside a Literal in an expression is not recognized as terminating the expression." (xslt rec 7.6.2) check-in: 1d7b74d309 user: rolf tags: trunk
00:44
Added a few more rsAddNodeFast and checkRsAddNode. check-in: d98de7757e user: rolf tags: trunk
2002-04-05
16:47
Bug fix for attribute value templates. Should now handle cases like attr="}}" and attr="{{" right. check-in: 18a313771b user: rolf tags: trunk
2002-04-02
00:36
Escape '\n' in Attribute Values while serializing, to fulfill the note in xslt rec. 7.1.3. check-in: bc13e06b37 user: rolf tags: trunk
2002-04-01
04:24
Closed a memory leak in not(). check-in: 1cf8371190 user: rolf tags: trunk
2002-03-31
03:27
Enhanced the useage of rsAddNodeFast(). Fixed a bug in Production(StepPattern) and closed a memory leak in xpathMatches(). check-in: b11313f86e user: rolf tags: trunk
03:22
Closed a memory leak in xsltFreeStats(). Changed xsl:element handling (work in progress). check-in: 5c2eaabf66 user: rolf tags: trunk
2002-03-25
13:56
Initial check in. check-in: eb5288f209 user: rolf tags: trunk
01:29
Closed some memory leaks. Fixed a bug in // handling. A bit code cleanup. check-in: 36b80d4fb1 user: rolf tags: trunk
01:28
Closed a cuple of memory leaks. A bit Code cleanup. check-in: 42fdb62c0c user: rolf tags: trunk
2002-03-23
00:31
Fix for multiple predicates on pattern steps. Patterns of the form foo[attr='c'][2] now work correct. check-in: a4ded5b2c6 user: rolf tags: trunk
2002-03-22
00:14
Removed some overseen debugging code. check-in: 88276c97b8 user: rolf tags: trunk
2002-03-21
01:47
Collected the various nodeSet Result types into "nodeSetResult" (there still exists a seperate emptyResult type). Reworked xpathEvalStep. Fixed memory leak in xpathMatches, added rsAddNodeFast(), if it's known for sure, that the node to add isn't already in the nodeSet. check-in: 0e6533ac59 user: rolf tags: trunk
2002-03-16
13:06
Optimised xsl:sort: the string or numeric value of the nodes to sort are computed only once and cached for further comparisons in the sort process. check-in: f00c8a6439 user: rolf tags: trunk
2002-03-14
13:28
Fixed the Fix of the last commit. check-in: cbe771bf12 user: rolf tags: trunk
01:27
Minimal Fix: doubled namespaceURI method corrected to prefix check-in: 7b7e247135 user: rolf tags: trunk
2002-03-10
01:21
Added documentation for the new dom command method createDocumentNS. check-in: 2fa449f347 user: rolf tags: trunk
01:20
Added tests for [dom createDocument] and the new [dom createDocumentNS]. Documented a bug (test dom-3.1). check-in: 473353233e user: rolf tags: trunk
01:17
Added method createDocumentNS to the dom command. check-in: ca9c194567 user: rolf tags: trunk
01:16
Added support for [dom createDocumentNS]. Added tests for correctness of document Element tag name. check-in: bd23419f09 user: rolf tags: trunk
01:14
Introduced distinction between XML Name and XML NC Name. check-in: 44b0ba5fd3 user: rolf tags: trunk
2002-03-08
23:49
Fixed small, obvious bug in DBG code. check-in: 6c5f787294 user: rolf tags: trunk
2002-03-07
22:16
Improved recognition of REALs (of the form .123). Made variable name recognition UTF-8 save. Freeze of actual state, befor feeding stuff to Jochen. check-in: 916956c205 user: rolf tags: trunk
22:11
Freeze of actual state, befor feeding stuff to Jochen. check-in: cf9a256595 user: rolf tags: trunk
22:09
Added infrastructur to be able to do NCNAME tests. Freeze of actual state, befor feeding stuff to Jochen. check-in: f180b1c8a2 user: rolf tags: trunk
22:04
Freeze of actual state, befor feeding stuff to Jochen. check-in: 80cde7bfa4 user: rolf tags: trunk
2002-03-05
16:15
Dummy config.h needed for expat. check-in: 687ef565b9 user: rolf tags: trunk
2002-03-03
20:08
Improved detection of improper stylesheets: now triggers error if any mandatory attribute of a xsl element is missing. The mandatory attributes are: xsl:stylesheet: version; xsl:transform: version; xsl:include: href; xsl:import: href; xsl:strip-space: elements; xsl:preserve-space: elements; xsl:call-template: name; xsl:namespace-alias: stylesheet-prefix result-prefix; xsl:element: name; xsl:attribute: name; xsl:attribute-set: name; xsl:processing-instruction: name; xsl:value-of: select; xsl:for-each: select; xsl:if: test; xsl:when: test; xsl:variable: name; xsl:param: name; xsl:copy-of: select; xsl:with-param: name; xsl:key: name match use check-in: 4b3d9115e0 user: rolf tags: trunk
01:02
tcltest test suite started. The expat related tests are mostly out of an older tclexpat test suite by Steve Ball (reviewed, bug fixed and adapted to tDOMs SAX interface). Plus a few tDOM specific tests. check-in: 8ba496bb01 user: rolf tags: trunk
2002-03-01
04:06
Improved detection of improper stylesheets: xsl:include xsl:import xsl:strip-space xsl:preserve-space xsl:apply-imports xsl:namespace-alias xsl:value-of xsl:number xsl:sort xsl:copy-of xsl:key xsl:decimal-format and xsl:output must be emtpy. check-in: 5350288274 user: rolf tags: trunk
01:30
No real code changes. Only to log additional change in 1.3: Fixed namespace overflow throu repeated (identical in prefix and uri) namespace declarations. check-in: 1ac62d594e user: rolf tags: trunk
01:18
Changed parsing. [dom parse ..] now uses Tcl_GetStringFromObj(), dom parse -channel now respects the encoding of the channel. check-in: 8cc106f218 user: rolf tags: trunk
01:16
Changed behavior of [<parserObj> parse <data>] and [<parserObj> parsechannel <Tcl channel>]. The method 'parse' now uses Tcl_GetStringFromObj(), the method 'parsechannel' now respects the encoding of the channel. check-in: 7f6fbcaa63 user: rolf tags: trunk
2002-02-28
00:57
Changed to new version number 0.7 check-in: 9874d60ad2 user: rolf tags: trunk
00:39
Added tcl coded xpath function element-avaliable. Changed function-avaliable accordingly. check-in: 5edd91ead4 user: rolf tags: trunk
2002-02-26
14:05
Updated the [load ...] to the new version number 0.7 check-in: 7d7e46a6ec user: rolf tags: trunk
14:04
Updated the [package provide] to the new version number 0.7 check-in: 5636300b8d user: rolf tags: trunk
2002-02-24
02:31
Fixed UTF-8 char byte length determination check-in: 9373983392 user: rolf tags: trunk
2002-02-23
01:13
Some code tweaking for a mostly warning free MS build check-in: 64a4af89f8 user: rolf tags: trunk
01:12
Removed some file noise check-in: 35a577fc36 user: rolf tags: trunk
01:08
Initial check in of win makefile. Tested on w2k with VC 6++ check-in: 0ce4795ef9 user: rolf tags: trunk
2002-02-22
01:05
tDOM0.7test with Jochens first set of patches Closed-Leaf check-in: 002482c973 user: rolf tags: demo, trunk
01:05
Initial revision check-in: 98cd4affb5 user: rolf tags: trunk