Personal tools
Skip to content. | Skip to navigation
D-Bus service for clients to discover and manipulate DLNA Digital Media Servers (DMSes).
dm.incrementalsearch is an efficient low level search execution engine. Its primary purpose is to incrementally perform searches involving and, or and not queries over ZODB.BTrees. Incrementally means here that hits are determined one at a time: the first hit, then the second hit, then the third hit, etc. Therefore, the first few hits can be determined extremely fast. But even if all hits need to be determined, the incremental execution of subqueries can lead to speedups of several orders for some query types (especially those dominated by specific and queries). Queries involving large or subqueries are difficult to optimize in the standard way. But often they can be replaced by incremental filtering. With this technique, a subquery is removed from the original search, the modified search executed and the result filtered by the removed subquery. incrementalsearch supports incremental filtering and thereby can again gain serveral orders of speedup for otherwise difficult to treat query types. The primary concept is that of an ISearch (incremental search). This is conceptionally a sorted list, computed incrementally (or lazily). The elements of this list are the ISearch's hits. The ISearch's keytype determines the type of the list elements. Currently supported are OBJECT (comparable Python objects), INT (Python 32 bit integers) and LONG (Python 64 bit integers).
Easier readable profile statistics -- especially for caller and callee analysis
"profile.Profile" improvement I started the implementation of this module in order to improve the profiler's calibration using statistical methods. Unfortunately, variance proved to be far too high and my calibration trials did not converge even for large sample sets. I had to abandon this approach. The code is still there (in case anyone would like to look what I have tried) but I do not use it. "pstats.Stats" improvement Beside the profiler's not so good calibration, I was not satisfied with "pstats" format especially for the caller and callee analysis and with its insistence to write to "stdout". Therefore, I derived a new class Stats from pstats.Stats and added new methods showStats, showCallers and showCallees which correspond to print_stats, print_callers and print_callees. They write to a file set with setOutputFile (default sys.stdout) and use a different output format. For showCallers and showCallees it is much more readable than the format used by print_callers and print_callees. I am not sure that this is also the case for the format used for showStats versus print_stats. When you are using "Stats.print_callers" or "Stats.print_callees" and find their output unreadable, then a switch to "dm.profile.Stats" may be profitable for you. The method showHeader outputs header information (with involved files, top level functions, number of calls and profiled time).
Support for object reuse with slight modifications
Utilities to reuse (slightly modified) oybjects in new contexts. Currently, there is a single utility: rebindFunction. It allows to reuse the code of a function while changing name, globals, default arguments, properties and/or names used.
This package provides support for SAML2 based on pyxb. pyxb (http://pypi.python.org/pypi/PyXB) generates a Python class collection for an XML schema and provides means to convert between associated Python instances and xml documents. It is used to generate and parse SAML2 messages. The package adds support for digital signatures and SAML2 bindings and metadata management.
This package contains a Cython (http://cython.org/) based bindung to Aleksey Sanin's XML security library ("http://www.aleksey.com/xmlsec") to be used together with lxml (http://lxml.de), the most popular Python binding to the Gnome XML library libxml2 (http://xmlsoft.org).