changelog tags files rss atom

shortlog for eventlet

navigate: (0) -100 -60 tip

4 weeks donovan 0.8pre
4 weeks donovan Added tag 0.7 for changeset f4a701d88d57
4 weeks donovan prepare for 0.7
4 weeks donovan merge from http://freehg.org/u/whichlinden/eventlet/
4 weeks rdw __init__.py restored
4 weeks which [svn r140] Fixed wsgi.py to capitalize headers, fixed printing of ConnectionError exceptions when raised by FileScheme, cleaned up merge messup in coros.
4 weeks which [svn r133] Minor adjustment to be 2.3-compatible.
4 weeks donovan [svn r121] Fix a bug in log_message, and fix a really stupid bug with where I call accept()
5 weeks rdw Heh, double-fix undone for _do_timeout. Discovered test_trampoline_timeout only works if it's the first test.
5 weeks rdw Merge latest from donovanpreston.com.
5 weeks rdw Fix import bustedness.
5 weeks rdw Merge r122:142 from svn trunk.
5 weeks rdw Merge
6 weeks which [svn r142] Fix for EVT-4: eventlet's https client is unable to communicate with it's https server. This turned out to be a combination of two factors: 1) the ssl server wasn't properly shutting down its socket before closing it, and 2) the client wasn't properly ignoring the error that this condition generates.
6 weeks which [svn r141] Added broken test case for SSL sockets, with supporting test server certs (self-signed, natch).
7 weeks which [svn r135] Trampoline timeout works now.
7 weeks which [svn r134] Fixes for some interesting issues discovered only after our count of unit tests using database connections crested 100.
2 months which [svn r132] Timeout calls previously didn't work because the number of arguments passed in to schedule_call didn't match the signature of the _do_timout callback.
2 months which [svn r131] Fix for some SwitchingToDeadGreenlet or TypeError: '_socketobject' object is not iterable exceptions: Passing both read=True and write=True to trampoline would cause the fd to select as both readable and writable in some error conditions. This would cause the greenlet to be spuriously resumed a second time. (commit comment stolen from donovan)
2 months nat [svn r128] With rdw's help, introduce an HttpStreamSuite subclass of HttpSuite whose
2 months nat [svn r127] Tidy up the implementation of CoroutinePool.wait_all() by overriding
2 months nat [svn r126] Add CoroutinePool.generate_results(function, iterable) method with signature
2 months nat [svn r125] Introduce new CoroutinePool methods:
2 months nat [svn r124] Publish greenlib.greenlet_dict() as api.local_dict() to provide a consistent
2 months which [svn r123] Fixed infinite redirect problem that Nat discovered.
5 weeks rdw Ignore more
5 weeks rdw Merge
5 weeks rdw Merge
5 weeks rdw Merge
3 months which [svn r122] Enable transparent 3xx redirection for GETs. Redirect loops are avoided by the use of the max_retries parameter. If this works out pretty well, we'll consider doing the same for other methods.
3 months q [svn r111] add ready() method to event class, so you can tell in advance whether or not wait() is going to block.
4 months seeping [svn r110] Added exception type for 408 (RequestTimeout), so that we can catch it
4 months kartic [svn r109] Added flush to NullSTDOut
4 months kartic [svn r108] DEV-6119 Backbone Traceback re: SysCallError: (104, 'Connection reset by peer') in indra.log
4 months donovan [svn r107] wrap_select_with_coroutine_select: This is to cooperate with people who are trying to do blocking
5 months which [svn r106] Cooked up a unit test for my earlier commit to coros.py, that exercises the flaw and demonstrates that it's solved.
5 months which [svn r105] Minor tweak to account for the fact that only some responses can be parsed into tracebacks in this way.
5 months seeping [svn r104] Added exception for response code 413: RequestEntityTooLarge.
5 months which [svn r103] Discovered some cases that could cause coroutine pools to leak coroutines. Added a try/finally to catch and fix these.
5 months seeping [svn r102] added exception classes for all the HTTP error codes needed to support CHTTP's exception semantics.
5 months which [svn r101] Actor is extended to be able to process multiple messages concurrently. The default is 1, naturally. The tests had to be changed substantially because the additional handoff associated with using the coroutine pool made single api.sleep(0) insufficient to force an execution of the received method. Also switched to a deque in pools.py since that maps more closely to the problem domain. Reviewed by jonathan and donovan.
5 months which [svn r100] Small tweaks to get saranwrap tests working on the Mac, mostly relating to changes in the behavior of copy and pickle modules between 2.3 and 2.5. It was bugging me.
5 months which [svn r99] Minor change to make doctest work on both 2.3 and 2.5.
5 months which [svn r98] Coroutine pools can now safely self-execute in the trivial case. The unit test decsribes the problem most succinctly. It turns out to be rather difficult to avoid inadvertently reentering a coroutine pool in code of moderate complexity, so it's better to simply make it safe to do so. Reviewed by Donovan.
5 months which [svn r97] Fix for reason phrase persisting across http/1.1 requests.
5 months which [svn r96] A batch of exception-related cleanup and fixes.
6 months donovan [svn r95] Remote tracebacks in https are now printed nicely on the client side
6 months which [svn r94] Reverted changes from r93 to httpd.py. While the tid is a useful concept, we should come up with a better design/implementation at some point in the future.
6 months seeping [svn r93] "nonblocking" support in tpool, to be exploited by dbmgr, so that
6 months which [svn r92] Argh, though I could get away with a last-minute pre-commit change but typo'd it. Never again\!
6 months which [svn r91] http://jira.secondlife.com/browse/EVT-13: Timer cleanup behavior is inconsistent between HTTP/1.1 and HTTP/1.0. This involved capturing a traceback on timer creation (based on a debug variable -- it'd be expensive to do that all the time), and adding timer cleanup code at the end of the handle method.
6 months which [svn r90] http://jira.secondlife.com/browse/EVT-14: Strange 400 error when using events and POST over HTTP/1.1. Turned out the problem was in the implementation of get_arg.
6 months sabin [svn r89] Changes made to allow dynamic resizing of the coroutine pool. Reviewed by Donovan.
6 months which [svn r88] Refactored the body-parsing logic that used to be in read_body() into parsed_body() and added some docs. This means that if you were calling read_body expecting to get anything but a string out, you should change to parsed_body() instead. Refactored tests.py a little bit.
6 months nat [svn r87] Fix bug in wrapped_fd.readuntil() (used by readline() et al.) in which a
6 months which [svn r86] Quieted noisy tests.
6 months which [svn r85] Minor bug in refactoring prevented db_pool_test from being run standalone on systems with MySQLdb installed. Also commented out the ab test cause finding the ab binary was not working most of the time, and when it was, it was spewing tons of output all over the place.
6 months which [svn r84] More documentation and doctests in coros.py. Turned tests.py into a utility that runs all eventlet tests that it can find. I'm much more likely to run all the tests if there's one command to do so. Refactored some of the tests slightly to play nicer with that style of test running.
7 months which [svn r83] Reflowed readme and put in references to eventletdev in the code.
7 months donovan [svn r82] Add max_http_version to httpd.server to allow specifying either http 1.0 or 1.1

navigate: (0) -100 -60 tip