File python-falcon.changes of Package python-falcon
-------------------------------------------------------------------
Mon May 4 17:49:09 UTC 2015 - benoit.monin@gmx.fr
- update to version 0.2:
* Since 0.1 we've added proper RTD docs to make it easier for
everyone to get started with the framework. Over time we will
continue adding content, and we would love your help!
* Falcon now supports "wsgi.filewrapper". You can assign any
file-like object to resp.stream and Falcon will use
"wsgi.filewrapper" to more efficiently pipe the data to the
WSGI server.
* Support was added for automatically parsing requests containing
"application/x-www-form-urlencoded" content. Form fields are
now folded into req.params.
* Custom Request and Response classes are now supported. You can
specify custom types when instantiating falcon.API.
* A new middleware feature was added to the framework. Middleware
deprecates global hooks, and we encourage everyone to migrate
as soon as possible.
* A general-purpose dict attribute was added to Request.
Middleware, hooks, and responders can now use req.context to
share contextual information about the current request.
* A new method, append_header, was added to falcon.API to allow
setting multiple values for the same header using comma
separation.
* A new "resource" attribute was added to hooks. Old hooks that
do not accept this new attribute are shimmed so that they will
continue to function.
* Error response bodies now support XML in addition to JSON. In
addition, the HTTPError serialization code was refactored to
make it easier to implement a custom error serializer.
* A new method, "set_error_serializer" was added to falcon.API.
You can use this method to override Falcon's default HTTPError
serializer if you need to support custom media types.
* Falcon's testing base class, testing.TestBase was improved to
facilitate Py3k testing.
* An "add_link" method was added to the Response class. Apps can
use this method to add one or more Link header values to a
response.
* Added two new properties, req.host and req.subdomain, to make
it easier to get at the hostname info in the request.
* Allow a wider variety of characters to be used in query string
params.
* Internal APIs have been refactored to allow overriding the
default routing mechanism. Further modularization is planned
for the next release (0.3).
* Changed req.get_param so that it behaves the same whether a
list was specified in the query string using the HTML form
style (in which each element is listed in a separate 'key=val'
field) or in the more compact API style (in which each element
is comma-separated and assigned to a single param instance, as
in 'key=val1,val2,val3')
* Added a convenience method, set_stream(...), to the Response
class for setting the stream and its length at the same time,
which should help people not forget to set both (and save a few
keystrokes along the way).
* Added several new error classes, including
HTTPRequestEntityTooLarge, HTTPInvalidParam, HTTPMissingParam,
HTTPInvalidHeader and HTTPMissingHeader.
* Python 3.4 is now fully supported.
* Various minor performance improvements
* The deprecated util.misc.percent_escape and
util.misc.percent_unescape functions were removed. Please use
the functions in the util.uri module instead.
* The deprecated function, API.set_default_route, was removed.
Please use sinks instead.
* HTTPRangeNotSatisfiable no longer accepts a media_type
parameter.
* When using the comma-delimited list convention,
req.get_param_as_list(...) will no longer insert placeholders,
using the None type, for empty elements.
* Ensure 100% test coverage and fix any bugs identified in the
process.
* Fix not recognizing the "bytes=" prefix in Range headers.
* Make HTTPNotFound and HTTPMethodNotAllowed fully compliant,
according to RFC 7231.
* Fixed the default on_options responder causing a Cython type
error.
* URI template strings can now be of type unicode under Python 2.
* When SCRIPT_NAME is not present in the WSGI environ, return an
empty string for the req.app property.
* Global "after" hooks will now be executed even when a responder
raises an error.
* Fixed several minor issues regarding testing.create_environ(...)
* Work around a wsgiref quirk, where if no content-length header
is submitted by the client, wsgiref will set the value of that
header to an empty string in the WSGI environ.
* Resolved an issue causing several source files to not be
Cythonized.
* Docstrings have been edited for clarity and correctness.
- additional changes from version 0.1.10:
* SCRIPT_NAME may not always be present in the WSGI environment,
so treat it as an empty string if not present.
- additional changes from version 0.1.9:
* Addressed style issues reported by the latest pyflakes version
* Fixed body not being decoded from UTF-8 in HTTPError tests
* Remove unnecessary ordereddict requirement on Python 2.6
- additional changes from version 0.1.8:
* srmock.headers have been normalized such that header names are
always lowercase. This was done to make tests that rely on
srmock less fragile.
* Falcon now sends response headers as all lower-case ala node.js.
* The 'scheme' argument to HTTPUnauthorized can no longer be
passed positionally; it must be a named argument.
* You can no longer overload a single resource class to respond
to multiple routes that differ by URI template params.
* srmock.headers_dict is now implemented using a case-insensitive
dict
* Per RFC 3986, Falcon now decodes escaped characters in the
query string, plus convert '+' -> ' '. Also, Falcon now decodes
multi-byte UTF-8 sequences after they have been unescaped.
* Custom error handlers can be registered via a new
API.add_error_handler method.
* Support for "request sinks" was added to falcon.API.
* uri module added to falcon.util which includes utilities for
encoding and decoding URIs, as well as parsing a query string
into a dict.
* Subsequent calls to req.uri are now faster since the property
now clones a cached dict instead of building a new one from
scratch each time.
* falcon.util now includes a case-insensitive dict borrowed from
the Requests library.
* Misc. performance optimizations to offset the impact of
supporting case-sensitive headers and rigorous URI
encoding/decoding.
* Py33 performance improvements
- add python-ddt and python-yaml as BuildRequires for the tests
- remove unwanted shebang of bench.py
- add -fno-strict-aliasing to CFLAGS to avoid compiler warnings
- pass -q to test to avoid spamming the build log
- remove AUTHORS CHANGES.md LICENSE: no provided anymore
- remove README.md: its content is identical to README.rst
-------------------------------------------------------------------
Mon Sep 30 15:31:12 UTC 2013 - speilicke@suse.com
- Fix description
-------------------------------------------------------------------
Tue Sep 17 08:49:17 UTC 2013 - dmueller@suse.com
- update to 0.1.7:
* req.get_params_as_list now inserts None as a placeholder for missing
elements, and returns None all by itself if the param is present in the
query string, but has an empty string as its value. (kgriffs)
-------------------------------------------------------------------
Mon Sep 9 09:22:07 UTC 2013 - speilicke@suse.com
- Add python-ordereddict test build requirement for SLE_11_SP3
-------------------------------------------------------------------
Mon Sep 9 08:24:35 UTC 2013 - speilicke@suse.com
- Initial version
