.. contents:: Contents ======================= Kid 0.9.6 Release Notes ======================= Enhancements ============ The ``format`` parameter can now be set to ``'straight'`` if you do not want any output formatting at all (the default format removes empty lines). Bug Fixes ========= These bugs in the last version have been fixed: 1. The Template ``_filters`` list is now an instance attribute (otherwise TurboKid may expand that list endlessly). 2. In some situations Kid printed misleading error tracebacks. 3. There was a small problem with indentation formatting. 4. The representation of the Format class was not correct. ======================= Kid 0.9.5 Release Notes ======================= Enhancements ============ Error messages -------------- Error reporting has been taken a step further: Kid now tries to track down the location of errors to the original Kid template file instead of merely printing the error location in the compiled template module. This has been one of the most urgently requested features. ElementTree not required any more --------------------------------- The dependency on the ElementTree package has been removed completely. Auto recompilation ------------------ Kid 0.9.3 templates needed to be recompiled in order to run with newer versions. This recompilation is now done automatically in the background whenever there is a new version, in order to ensure compatibility and to benefit from improvements in template creation in newer versions. Changes ======= The old Kid namespace URL ``http://naeblis.cx/ns/kid#`` is not supported any more. Please use the new URL ``http://purl.org/kid/ns#`` which has been introduced already in Kid 0.6. Bug Fixes ========= These bugs in the last version have been fixed: 1. Combination of both ``py:layout`` and ``py:extends`` caused a problem. 2. Passing parameters to functions with default parameters which were defined with ``py:def`` did not work properly. 3. The XML encoding declaration in the template was not evaluated. ======================= Kid 0.9.4 Release Notes ======================= Enhancements ============ Error messages -------------- XML errors from the Expat parser are now shown along with the corresponding line of the erroneous XML code and the exact position of the error. Kid is now much less picky about ``py:extends`` and ``py:layout`` expressions and will output much better error messages if they cause an error. Kid will now also check that you dont' use a Template parameter that conflicts with the name of a member of the BaseTemplate class. Output Format Control --------------------- The serialization methods now take an additional attribute ``format`` that must be an instance of the new Format class or a string referring to one of several predefined output formats. This class gives you more fine control over how the text content is serialized. You can do things like removing redundant whitespace, adding indentation, word wrapping, using named entities and using typographic characters for quotes etc. You can also plug-in your own text filters here. The chapter on Formatting in the User's Guide lists all available parameters. The old filter methods of the Serializer class have been replaced by a more powerful general method incorporating the specified format. HTML Serializer --------------- As discussed on the mailing list, the HTML serializer now produces tags in lowercase. You can change this behavior with the transpose attribute or using the output method 'HTML' instead of 'html'. Additional HTML output methods have been defined corresponding to Mozilla's "full standards mode", "almost standards mode" and "quirks mode." XHTML Serializer ---------------- The XHTML Serializer now also injects a meta tag with the content type at the top of the head section. This had been only done by the HTMLSerializer before. The content type is not injected if such a meta tag already exists. This automatism can be controlled with the ``inject_type`` parameter. XML Comments ------------ The rule when variable substitution is applied to XML comments has been slightly changed, as explained in the Language Specification. Additional parameters for parsing templates ------------------------------------------- The ``load_template()`` function now takes additional parameters ``entity_map`` and ``exec_module``. This lets you choose a different XML entity map to be used when parsing the template, and gives you more control over the way in which the compiled template code is executed in the newly created module. The ``entity_map`` parameter can also be passed to some other functions such as ``XML()`` and ``compile_file()``. Full compatibility with Python 2.5 and Python eggs -------------------------------------------------- The enable_import mechanism has been completely rewritten to support both Python 2.5 and Python eggs. The old method based on `ihooks` is not used any more since `ihooks` is not compatible with Python eggs; and the old method based on `path_hooks` is not used any more since in Python 2.5, importers installed via `path_hooks` do not fall back to the built-in import mechanism any more, breaking the previous implementation. The new implementation allows enabling imports from specified paths as well as imports from `sys.path` which is now implemented via `meta_path`. Python versions older than Python 2.3 are not supported by Kid any more. Changes ======= Due to optimizations in `kid.template_util`, Kid 0.9.3 templates need to be recompiled in order to run with Kid 0.9.4. The ``'html'`` output method now produces lower-case tags. If you want upper-case tags, you have to use the ``'HTML'`` method now. The `kid.parser` and `kid.pull` modules have been renamed `kid.codewriter` and `kid.parser` respectively (in anticipation of a parsing subsystem). Bug Fixes ========= A couple of bugs that had been reported as trac tickets have been fixed. Using layout templates which are extended by another template did not work properly (as reported on the TurboGears mailing list); this has been fixed. More unit tests have been added. The testing machinery now supports both `py.test` and `nose`, or you can simply use the ``run_tests.py`` script. ======================= Kid 0.9.3 Release Notes ======================= 1. Re-applied a patch from ticket [66] that fix a bug where comments caused errors when in base templates. 2. Changed all of the lesscode.org links into kid-templating.org 3. Added and updated a few tests 4. Removed the NamespaceStack.set method and made NamespaceStack.pop return the deleted value. 5. Set balanced blocks to be off by default. 6. Updated the parser to better handle interpolation of non-string types in comments. Reported in #182 ======================= Kid 0.9.2 Release Notes ======================= Enhancements ============ 1. Updated to current version of ez_setup.py. 2. Improved importer.py, resolving tickets #103 (FutureWarnings) and #137 (using new import hooks). 3. The testing code can now figure out what testing modules to run dynamically. In addition, the code also determines which functions are tests dynamically. Tests that need pylib are skipped for those that don't have it. If you run 'python test_kid.py' now you should be seeing more tests executed. 4. Removed the revision history from the language and and instead include a pointer to the Release Notes. API Changes ----------- 1. Allow the kid command to accept XML piped into it's stdin when '-' is used as the filename. 2. Patch from #143. The load_template() function accepts an 'ns' keyword argument to pre-populate the template module namespace with global variables. Thanks! 3. Created an API to replace the various ways that configuration options are currently set. 4. Allow the XML function to take a new keyword parameter (xmlns) that sets the default namespace for a fragment. Add Support For Python 2.5 -------------------------- 1. Added support for xml.etree, which is the ElementTree packaged with Python 2.5. Running 'make test' only checks xml.etree currently. 2. The __future__ imports have been moved to the top of the module to play nicely in Python 2.5. I have also added 2.5 to the makefile so it will be tested before each release. 3. makefile regression test includes Python 2.5. ======================= Kid 0.9.1 Release Notes ======================= Bug Fixes ========= Layout Templates ---------------- The parameters passed to a template with a py:layout were not visible in named template functions or match templates. A small bug existed in Python 2.3 where the dict.update() method was being called incorrectly. Python 2.4 allows a list of tuples to be passed to update(), whereas Python 2.3 does not. py:match -------- The logic to apply the template matches has been reworked. This was due to the discovery of some odd behavior when using multiple template inheritence. Enhancements ============ Layout Templates ---------------- A template type or name can now be dynamically passed into a template to be used as the layout template. ===================== Kid 0.9 Release Notes ===================== Language and API Changes ======================== Layout Templates ---------------- There is a new feature in Kid for a template to specify a layout template to which match templates, named template definitions, and template parameters will be applied. This is useful for applying a generic set of headers, menus, footers, etc. to a many pages without duplicating large amounts of code in each page. More information and examples can be found in the `py:layout`__ section of the Language Specification. .. __: language.html#py:layout Convenience Functions --------------------- Kid Template instances now provide convenience funcitons ``defined(name)`` and ``value_of(name)``. Invisible Comments ------------------ XML comments starting with a ``!`` character will not appear in the serialized output of a template. Enhancements ============ Command line scripts now work on Windows ---------------------------------------- On Windows, the ``kid`` and ``kidc`` console commands can now be used as conveniently as on Unix, since the Kid installer creates ``kid.exe`` and ``kidc.exe`` launchers in the Python Scripts directory. ===================== Kid 0.6 Release Notes ===================== There has been significant change in version 0.6. This includes enhancements and modifications to the template language and python interface. Language and API Changes ======================== The following changes are likely to impact existing code and templates. Where possible, we have tried to maintain backward compatibility but that wasn't possible in all cases. The `Upgrade Script`_ can be used to bring 0.5 templates up to 0.6 syntax. Kid Namespace Change -------------------- The Kid namespace has changed from ``http://naeblis.cx/ns/kid#`` to ``http://purl.org/kid/ns#``. The naeblis.cx domain is privately owned and could expire some time in the future. purl.org is a system for establishing and maintaining "persistent" URIs. A temporary hack has been put in place to substitute references to the old namespace URI with the new namespace URI. A warning is output when this occurs. This will be removed in a couple of months so it is recommended that templates be upgraded as soon as possible. ``py:omit`` is now ``py:strip`` ------------------------------- Due to initial confusion many experienced with the name ``py:omit``, it has been renamed ``py:strip``. The term "omit" was often read as "omit the element and all descendants". The new term "strip" seems to better indicate the semantic: "strip the start and end tag but process descendants." Python Expression Substitution Syntax ------------------------------------- The syntax of brace expansions has been modified match more closely with existing Python substitution syntax. In 0.5 python expressions enclosed in curly braces ({}) were evaluated and their results substituted. In 0.6, the rules have changed as follows: 1. ``$$`` is an escape; it is replaced with a single $. 2. ``$name`` substitutes a variable value. 3. ``${expr}`` substitutes the result of evaluating any python expression. See `Python Expression Substitution`__ in the Language Reference. .. __: language.html#python-expression-substitution-expr .. warning: The expression substitution **is not** backward compatible. If you use the old-style brace-expansion, you will need to upgrade your templates for Kid 0.6. Enhancements ============ cElementTree Support -------------------- Kid now uses cElementTree if it is available. Preliminary tests show moderate performance increases. In most cases, we're seeing template parse and execution time increase by about 15%. The poor increase (relative to other cET/ET numbers) is due to the fact that we're not using cElementTree's native parser as it doesn't support comments or processing instructions. The plan is to lobby the effbot organization to add these features (hint, hint: send patches) so that we can get the huge increases people are seeing elsewhere. Kid automatically determines whether cElementTree is available and uses it if so. If cElementTree is not available, Kid falls back on Python ElementTree. If you want to turn off use of cElementTree, you can set the environment variable ``KID_NOCET`` to 1. Death of Comment Wart --------------------- In versions of Kid prior to 0.6, the first line of an embedded Python code block had to be a Python comment (#). This was due to Python's whitespace semantics. Christoph determined a process for establishing the correct indent levels without requiring a comment as the first line. Starting in Kid 0.6, a comment is no longer required to be the first line in a ```` processing instruction. It is also possible to have single line code blocks:: Improved Template API --------------------- The Python interfaces have been reworked significantly and now are very similar to Cheetah's. There are two preferred methods for accessing a template. The ``Template`` Class ~~~~~~~~~~~~~~~~~~~~~~ The first method existed in 0.5 but was not documented well. If you have enabled the kid import hooks, then you can import a template and create an instance of the template by accessing the ``Template`` class exposed by the module:: import kid ; kid.enable_import() import mytemplate template = mytemplate.Template(foo='bar', bling=1) print template.serialize() The primary difference from 0.5 is that template variables are passed to the ``Template`` constructor instead of to the individual execution methods (``serialize``, ``generate``, ``write``, ``pull``). It is also possible to set template variables after the template instance is created by simply assigning to template object instance:: template = mytemplate.Template() template.foo = 'bar' template.bling = 1 print str(template) Here we see another small addition: template instances implement ``__str__`` and ``__unicode__`` built-ins. These methods are equivalent to calling ``serialize(encoding='utf-8')`` and ``serialize(encoding='utf-16')``, respectively. The ``kid.Template`` function ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The ``kid.Template`` function works much like ``Template`` class constructors but takes an additional parameter that allows the template to be loaded from a file, string, or module name. It is sometimes easier to manage templates as files on disk rather than as python modules. Example:: from kid import Template # create a template from file template = Template(file='mytemplate.kid', foo='bar', bling=1) # create a template from string template = Template(source="

${foo}

", foo='bar') # create a template from a python module name template = Template(name='templates.mytemplate', foo='bar') This last form is sometimes useful because it doesn't require the kid import hook to be enabled and it also allows template names to be specified at run-time. See `kid.Template function`__ in the User's Guide for more info. .. __: guide.html#template-function Match Templates / Filters ------------------------- Match Templates are a cross between XSLT's match templates and JSP tag libraries. They allow a set of filters to be put in place that matches infoset items generated by a template so that output can be modified. While match templates provide a general purpose mechanism for transforming XML content, it is especially useful in a couple of situations which have driven the design: 1. Creating tag libraries that inject new tags into an XML vocabulary. 2. Applying headers/footers without inserting place-holders into the source document/template. See `Match Templates`__ in the Language Reference for more information. .. __: language.html#match-templates Template Inheritance -------------------- Templates now support multiple inheritance of template functions (``py:def``) and match templates (``py:match``). A template indicates that it extends one or more other templates by setting the ``py:extends`` attribute on the root element::