Tue Jun 6 17:16:33 CEST 2006 Daniel Veillard * configure.in libxslt.spec.in doc//*: preparing release of 1.1.17 Tue Jun 6 11:44:34 CEST 2006 Kasimier Buchcik * libxslt/attributes.c: Eliminated a tiny difference wrt to the old behaviour in the naming of newly generated ns-decls in xsltAttributeInternal(). Thu Jun 1 15:06:31 CEST 2006 Daniel Veillard * configure.in: fix a problem in Python detection Thu Jun 1 13:58:19 CEST 2006 Kasimier Buchcik * libxslt/attributes.c libxslt/variables.c libxslt/transform.c libxslt/xslt.c: Next step in the refactored code: enhanced xsl:attribute; enhanced xsltCopyProp. Added the creation of an XPath cache in xsltNewTransformContext(). Wed May 31 22:32:44 CEST 2006 Kasimier Buchcik * libxslt/transform.c: Fixed a difference in processing of xsl:value-of and xsl:text wrt to merging of strings of output-escaped text-nodes. This difference made the optimized string-merging mechanism run out of sync, which led to segfaults in subsequent string reallocations. See bug #343411, reported by Grzegorz Kaczor. * xslt.c: Added check for ctxt->internalized when we parse xsl:text and internalize the strings. Mon May 22 10:32:57 CEST 2006 Kasimier Buchcik * libxslt/attributes.c libxslt/documents.c libxslt/extensions.c libxslt/keys.c libxslt/pattern.c libxslt/preproc.c libxslt/templates.c libxslt/transform.c libxslt/variables.c libxslt/xslt.c libxslt/xsltInternals.h: Next tiny step of refactoring - mostly bug fixes and cosmetic changes. Changes outside of the refactored code: 1) Optimized xsl:attribute if the content consists of just 1 text node. 2) Optimized computation of xsl:key. The keys will now be computed for a specific document not until the first call of a key() function; here only the keys with the specific name used by key() are computed. This means that this now avoids computation of all keys for all loaded input documents (even if no key() was called on them). One exception is the scenario where a key() is used in a template's match pattern; in this case all keys are computed for a document if there's a chance that a "keyed" template could match a node (this could still be optimized a bit). Mon May 15 22:32:13 CEST 2006 Kasimier Buchcik * libxslt/namespaces.c libxslt/attributes.c: Fixed bug #302020, reported by Thomas Blatter. Fri May 12 23:23:06 CEST 2006 Kasimier Buchcik * libxslt/documents.c libxslt/namespaces.c libxslt/preproc.c libxslt/transform.c libxslt/xslt.c libxslt/xsltInternals.h libxslt/xsltutils.c: Next step of refactoring. For more details see bug #341588. I applied the suggestion of Jerome Pesenti to the refactored (still IDFEDed out with XSLT_REFACTORED) code: The XPath compilation context (accessible via the compilation context) is now reused for compilation of expressions at compilation-time; this should reduce compilation time to 50% for avarage stylesheets. Thu May 11 22:12:22 CEST 2006 Kasimier Buchcik * libxslt/extensions.c: Changed a comment to indicate that a specific bug was already fixed. Fri May 5 23:10:47 CEST 2006 Kasimier Buchcik * libxslt/xsltInternals.h libxslt/attributes.c libxslt/documents.c libxslt/extensions.c libxslt/extensions.h libxslt/functions.c libxslt/imports.c libxslt/keys.c libxslt/preproc.c libxslt/transform.c libxslt/variables.c libxslt/xslt.c libxslt/xsltutils.c libxslt/xsltutils.h libexslt/functions.c: Next step of refactoring (plus some bug-fixes). For more details see #340780. Fri May 5 14:31:53 CEST 2006 Kasimier Buchcik * tests/exslt/common/node-set.5.out tests/exslt/sets/difference.1.out tests/exslt/functions/function.6.out: Reverting the changes; I'll rather change the processing stylesheet documents in a way that they will produce the same results for broken and stricter whitespace-stripping. Thu May 4 22:55:26 CEST 2006 Kasimier Buchcik * tests/exslt/common/node-set.5.out tests/exslt/sets/difference.1.out tests/exslt/functions/function.6.out: Changed regression test results (bug #340684). Those will now produce regression errors; so we need to fix the processor.