repoze.profile Aggregate profiling for WSGI requests repoze.profile README ===================== This package provides a WSGI middleware component which aggregates profiling data across *all* requests to the WSGI application. It provides a web GUI for viewing profiling data. Installation ------------ Install using setuptools, e.g. (within a virtualenv):: $ easy_install repoze.profile Configuration via Python ------------------------ Wire up the middleware in your application:: from repoze.profile.profiler import AccumulatingProfileMiddleware middleware = AccumulatingProfileMiddleware( app, log_filename='/foo/bar.log', cachegrind_filename='/foo/cachegrind.out.bar', discard_first_request=True, flush_at_shutdown=True, path='/__profile__' ) The configuration options are as follows:: - ``log_filename`` is the name of the file to which the accumulated profiler statistics are logged. - ``cachegrind_filename`` is the optional name of the file to which the accumulated profiler statistics are logged in the KCachegrind format. - If ``discard_first_request`` to true (the default), then the middleware discards the statistics for the first request: the rationale is that there are a bunch of lazy / "first time" initializations which distort measurement of the application's normal performance. - If ``flush_at_shutdown`` is true (the default), profiling data will be deleted when the middleware instance disappears (via its __del__). If it's false, profiling data will not be deleted. - ``path`` is the URL path to the profiler UI. It defaults to ``/__profile__``. Once you have some profiling data, you can visit ``path`` in your browser to see a user interface displaying profiling statistics (e.g. ``http://localhost:8080/__profile__``). Configuration via Paste ----------------------- Wire the middleware into a pipeline in your Paste configuration, for example:: [filter:profile] use = egg:repoze.profile#profile log_filename = myapp.profile cachegrind_filename = cachegrind.out.myapp discard_first_request = true path = /__profile__ flush_at_shutdown = true ... [pipeline:main] pipeline = egg:Paste#cgitb egg:Paste#httpexceptions profile myapp Once you have some profiling data, you can visit ``path`` in your browser to see a user interface displaying profiling statistics. Reporting Bugs / Development Versions ------------------------------------- Visit http://bugs.repoze.org to report bugs. Visit http://svn.repoze.org to download development or tagged versions. Changelog ========= 1.0 (unreleased) ---------------- - Relax the pinned requirement on elementtree < 1.2.7. 0.9 (2009-05-10) ---------------- - Made the `pyprof2calltree` dependency conditional on the Python version. This package depends on Python >= 2.5. 0.8 (2009-02-25) ---------------- - Added optional support for directly writing out the profiling data in the KCacheGrind format. - Avoid a dependency on `elementtree` when used with Python 2.5 and later. In those Python versions we used the built-in xml.etree support. 0.7 (2009-02-08) ---------------- - ``discard_first_request = true`` did not work! - Added tests for ``discard_first_request`` and ``flush_at_shutdown``. - Converted CHANGES.txt to ReST. - Bump ez_setup.py version. 0.6 (2008-08-21) ---------------- - ``discard_first_request = false`` did not work. - Clearing the profile data from the user interface did not properly discard profiler state. 0.5 (2008-06-11) ---------------- - Initial PyPI release. 0.4 (2008-05-07) ---------------- - Remove dependency-link to http://dist.repoze.org to prevent easy_install from adding that to its search path. - Incorporated a patch from Alec Flett <alecf@metaweb.com> to uses ``cProfile``, if available, rather than pure Python ``profile``. - Bump ez_setup.py version. 0.3 (2008-02-20) ---------------- - Added compatibility with Python 2.5. - Made setup.py depend explicitly on ElementTree 1.2.6: meld needs it but meld isn't a setuptools package. 0.2 (2008-02-20) ---------------- - Added a browser UI. - Added a knob to control discard at shutdown. 0.1 (2008-02-08) ---------------- - Initial release. Agendaless Consulting 077af4d52387f179a077d47e9bb8c9583a170f6a 1.0