summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* subprocess: Fix patched functions with optional argumentssubprocessJakub Stasiak2016-07-112-1/+15
| | | | | | | | | | | | | | Since [1] we monkey patch subprocess module and it uncovered one issue - the previous patched_function implementation would produce functions with all optional arguments lacking their default values which made them required. This is an issue on Python 3 where there's at least one optional explicit argument in at least one of the functions (check_call), on Python 2 the function signature has *args and **kwargs so it wasn't an issue. This patch is contributed by Smarkets Limited. [1] 614a20462aebfe85a54ce35a7daaf1a7dbde44a7
* Provide green http without undesired side effectsJakub Stasiak2016-07-0812-42/+5773
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The Eventlet patcher and the way we were patching multi-level http package don't work well[1][2]. I spent a lot of time trying to make it work but in the end every solution I came up with was breaking something else and made the patching and providing green http even more complicated - I wouldn't envy anyone having to debug it in the future. After a lot of thinking I decided having our own copy of http with the necessary modifications applied seems like the most straightforward and the most reliable solution, even considering its downsides (we need to keep it up to date ourselves and the API won't 100 % match the regular http module API on older Python 3 versions as our bundled version is the most recent one and has bug fixes and extra features implemented). The code introduces by this commit comes from the following Python commit (development branch): commit 6251d66ba9a692d3adf5d2e6818b29ac44130787 Author: Xavier de Gaye <xdegaye@users.sourceforge.net> Date: 2016-06-15 11:35:29 +0200 Issue #26862: SYS_getdents64 does not need to be defined on android API 21. Changes to the original http package code involve: * Removing unnecessary import(s) * Replacing some regular imports with eventlet.green imports * Replacing fullmatch()[3] usage with match() so we stay Python 3.3 compatible I left urllib.parse imports intact as nothing there performs IO. Green httplib module is also modified because it used to import http.client using patcher which was breaking things the same way. A new dependency, enum-compat, is added to ensure that the enum module is present on Python 3.3 (the http package code comes the latest Python development branch and uses enum). [1] https://github.com/getsentry/raven-python/issues/703 [2] https://github.com/eventlet/eventlet/issues/316 [3] https://docs.python.org/3/library/re.html#re.fullmatch This patch is contributed by Smarkets Limited.
* greenio: makefile related pypy socket ref countingYashwardhan Singh2016-07-071-0/+5
| | | | | | | On PyPy `[Errno 9] Bad file descriptor` because `makefile dup/_drop` dance leaves socket with PyPy special reference counter zero, so it may be garbage collected. https://github.com/eventlet/eventlet/issues/318 (@temoto) maybe proper fix is to remove `dup` in `makefile()`.
* websocket: support Gunicorn environ['gunicorn.socket']pull-331catroot2016-07-062-2/+61
| | | | | | Now you able to upgrade connection with eventlet.websocket.WebSocketsWSGI when app runs under `gunicorn application:app -k eventlet` https://github.com/eventlet/eventlet/pull/331
* greenio: only trampoline when we blockDavid Szotten2016-07-013-2/+49
| | | https://github.com/eventlet/eventlet/pull/314
* tests: clean out unittest.main()Sergey Shepelev2016-06-016-44/+11
|
* hub: Python2 _obsolete() was copying dicts, use six.iteritemsSergey Shepelev2016-05-241-3/+3
|
* socket: family kwarg name compatibilityissue-319Sergey Shepelev2016-05-242-4/+10
| | | | | GreenSocket was (family_or_realsock=AF_INET, ...) which breaks code like socket(family=...) https://github.com/eventlet/eventlet/issues/319
* release script: doc build was breaking pypi uploadSergey Shepelev2016-05-241-0/+1
|
* Remove select.poll and improve subprocessJakub Stasiak2016-05-206-16/+37
| | | | | | | | | | | | | | | | | | | | | | | | green select: Delete unpatched poll once again https://github.com/eventlet/eventlet/pull/317 Previously attempted in f63165c, had to be reverted in 8ea9df6 because subprocess was failing after monkey patching. Turns out we haven't been monkey patching the subprocess module at all, this patch adds that in order for the tests to pass. This part is changed because otherwise Popen class instantiation would cause an infinite loop when monkey patching is applied: -subprocess_orig = __import__("subprocess") +subprocess_orig = patcher.original("subprocess") This patch is contributed by Smarkets Limited. * green subprocess: Provide green check_output This patch is contributed by Smarkets Limited.
* Add Python 3.5 to test envs; Thanks to Piët DelportSergey Shepelev2016-05-183-3/+9
|
* greendns tests: Work around patcher/green package weirdnessJakub Stasiak2016-05-181-6/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The issue can be demonstrated by running the following piece of code: # t.py from __future__ import print_function from eventlet.support import greendns import sys if sys.argv[1] == 'yes': import dns.resolver print(sys.argv[1], issubclass(greendns.HostsAnswer, greendns.dns.resolver.Answer)) The results: # Python 2.7.11 % python t.py yes yes False % python t.py no no True # Python 3.5.1 % python t.py yes yes False % python t.py no no True Interestingly enough this particular test issue was only affecting Python 3.5+ before 861d684. Why? * This issue appears to be caused by importing green version of a package being followed by importing a non-green version of the same package * When we run tests using nose it first imports the main tests module (tests/__init__.py) which imports eventlet, that imports eventlet.convenience and that then imports eventlet.green.sockt. * Before 861d684 on Python < 3.5 the eventlet.green.socket import mentioned above would fail to import greendns (because of an import cycle) so when running those tests greendns was only being correctly imported *after* the regular dns * Since 861d684 (or on Python 3.5+) the green socket module correctly imports greendns which means that when the regular dns subpackages are being imported in this test file greendns is already imported and the patching issue demonstrated by the code above is in effect The patching/greening weirdness is reported[1] now. Fixes https://github.com/eventlet/eventlet/issues/267 This patch is contributed by Smarkets Limited. [1] https://github.com/eventlet/eventlet/issues/316
* socket: Actually provide non-blocking DNS methodsJakub Stasiak2016-05-183-5/+32
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The green socket module seemed to have only blocking DNS resolution methods even with dnspython installed which is inconsistent with the documentation. This patch has a few consequences: * an import cycle is eliminated * if an import cycle reappears here it'll be visible Note: eliminating the import cycle revealed an issue related to monkey patching and the way we perform greendns tests (the test failures were already present on Python 3.5[1] as that version has some import cycle handling changes). The failures look like this: ====================================================================== FAIL: test_query_ans_types (tests.greendns_test.TestHostsResolver) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Users/kuba/projects/eventlet/tests/greendns_test.py", line 97, in test_query_ans_types assert isinstance(ans, greendns.dns.resolver.Answer) AssertionError ====================================================================== FAIL: test_query_unknown_no_raise (tests.greendns_test.TestHostsResolver) ---------------------------------------------------------------------- Traceback (most recent call last): File "/Users/kuba/projects/eventlet/tests/greendns_test.py", line 129, in test_query_unknown_no_raise assert isinstance(ans, greendns.dns.resolver.Answer) AssertionError This issue will be addressed in a separate commit. This patch is contributed by Smarkets Limited. [1] https://github.com/eventlet/eventlet/issues/267
* greendns: Reduce code redundancyJakub Stasiak2016-05-181-15/+13
| | | | | | | | This list of module dependencies are already present in the code three times and I intend to make a change that will require the same dependencies reused even more. This patch is contributed by Smarkets Limited.
* tests: Make greendns tests less repetitiveJakub Stasiak2016-05-181-80/+1
| | | | This patch is contributed by Smarkets Limited.
* v0.19.0 releasev0.19.0Sergey Shepelev2016-05-024-2/+14
|
* tox: don't cap pyopenssl to 0.13tox_pyopensslVictor Stinner2016-04-231-1/+2
| | | | | | | | | | | | | | | | | | | | | | pyopenssl 0.13 cannot be installed on my Fedora 23. It fails with: building 'OpenSSL.crypto' extension (...) OpenSSL/crypto/crl.c:6:23: erreur: static declaration of ‘X509_REVOKED_dup’ follows non-static declaration static X509_REVOKED * X509_REVOKED_dup(X509_REVOKED *orig) { ^ In file included from /usr/include/openssl/ssl.h:156:0, from OpenSSL/crypto/x509.h:17, from OpenSSL/crypto/crypto.h:30, from OpenSSL/crypto/crl.c:3: /usr/include/openssl/x509.h:751:15: note: previous declaration of ‘X509_REVOKED_dup’ was here X509_REVOKED *X509_REVOKED_dup(X509_REVOKED *rev); ^ error: command 'gcc' failed with exit status 1 The bug is known: https://github.com/pyca/pyopenssl/issues/276 The workaround is simple: use a more recent version.
* green.urllib2: missing patched ssl moduleCollin RM Stocks2016-04-224-1/+6
|
* patcher: patch existing threading locks; Thanks to Alexis LeeAlexis Lee2016-04-187-2/+177
| | | | | | | | | | | | | | | | | | In projects which dynamically determine whether to activate eventlet, it can be hard not to import a low level module like logging before eventlet. When logging is imported it initialises a threading.RLock which it uses to protect the logging configuration. If two greenthreads attempt to claim this lock, the second one will block the /native/ thread not just itself. As green systems usually only have one native thread, this will freeze the whole system. Search the GC for unsafe RLocks and replace their internal Lock with a safe one while monkey-patching. The tests pass, but were they to fail, the test process would never return. To deal with this, I've added a test dependency on subprocess32 which is a backport of the stdlib subprocess module from Python3. This offers a timeout option on Popen#communicate, which I've arbitrarily set at 30 seconds.
* Minor grammatical improvements and typo fixes to the docsSteven Erenst2016-03-033-7/+7
| | | | Signed-off-by: Steven Erenst <stevenerenst@gmail.com>
* tests: half-hearted fix sporadic zmq_test failures on TravisSergey Shepelev2016-03-031-59/+73
| | | | | | | | Real fix must work without any sleep(), even 0 delay. With all these sync events the code makes perfect sense and execution is properly organized. So I am 99% sure there is a problem in green/zmq. Hint for the person willing to take on this in future: send/recv seems not to switch to zmq trampolining
* wsgi: environ[headers_raw] tuple of unmodified name: value pairsSergey Shepelev2016-03-022-2/+15
| | | | https://github.com/eventlet/eventlet/pull/278
* ssl: IMPORTANT DoS FIX do_handshake_connect=False in server accept(); Thanks ↵ssl-handshakeSergey Shepelev2016-02-293-73/+107
| | | | | | | | | | | | | | | | to Garth Mollett Default is: accept - handshake (blocking) - return new connection Now: accept - return new connection, some time later handshake will be done implicitly Usual server code: while server.alive: conn, addr = listener.accept() server.pool.spawn(server.process, conn, addr) is vulnerable to a simple DoS attack, where evil client connects to HTTPS socket and does not perform handshake, thus blocking server in `accept()` so no other clients can be accepted.
* wsgi: document compatibility with python `logging`lbonn2016-02-221-1/+2
| | | | | Mention that `Logger` objects can be supplied to wsgi server instances (work done in #75)
* v0.18.4 releasev0.18.4Sergey Shepelev2016-02-173-2/+6
|
* wsgi: TCP_NODELAY -> TCP_QUICKACKSergey Shepelev2016-02-171-7/+7
| | | | | | | Also setsockopt(TCP_NODELAY) sometimes failed (UNIX sockets?). https://github.com/eventlet/eventlet/issues/301 https://news.ycombinator.com/item?id=10607422
* v0.18.3 releaseSergey Shepelev2016-02-135-12/+20
|
* wsgi: use buffered wfileSergey Shepelev2016-02-124-36/+56
| | | | | | | | Fix the root cause of makefile().writelines() data loss. https://github.com/eventlet/eventlet/issues/295 Also wsgi.log.write can break file-object API and not return number of bytes again. https://github.com/eventlet/eventlet/issues/296
* wsgi: TCP_NODELAY enabled by defaultSergey Shepelev2016-02-121-0/+9
|
* tpool: yield after setup() and before killall()Sergey Shepelev2016-02-121-7/+22
| | | | | Fixes surprising errors in artificial IO-dry scenarios (like tests). Won't hurt in real environment.
* little clean upSergey Shepelev2016-02-1224-583/+439
|
* build: automate release commitSergey Shepelev2016-02-112-14/+125
|
* v0.18.2 releasev0.18.2Sergey Shepelev2016-02-093-4/+8
|
* wsgi: Fix handling partial writes on Python 3partial-write-fix-2Jakub Stasiak2016-02-094-8/+88
| | | | | | | | | | | | | Closes https://github.com/eventlet/eventlet/issues/295 (in the wsgi module we use a custom writelines implementation now). Those write() calls might write only part of the data (and even if they don't - it's more readable to make sure all data is written explicitly). I changed the test code so that the write() implementation returns the number of characters logged and it cooperates nicely with writeall() now.
* wsgi tests: Remove unused importJakub Stasiak2016-02-091-1/+0
|
* v0.18.1 releasev0.18.1Sergey Shepelev2016-01-253-3/+7
|
* patcher: certain order of import subprocess and monkey_patch breaks ↵issue-290Sergey Shepelev2016-01-244-4/+31
| | | | | | .communicate() https://github.com/eventlet/eventlet/issues/290
* greenio: Fix "TypeError: an integer is required" in sendto()Jakub Stasiak2016-01-242-2/+17
| | | | | | | | | | | | | | | | | The sendto() interface as defined in Python documentation: socket.sendto(string, address) socket.sendto(string, flags, address) I didn't catch the fact that [1] broke this, this patch fixes it and add a sendto/recvfrom test to make sure it doesn't happen again (turns out we didn't have any). GitHub issue: https://github.com/eventlet/eventlet/issues/290 Fixes: bc4d1b5 - gh-274: Handle blocking I/O errors in GreenSocket [1] bc4d1b5d362e5baaeded35b1e339b9db08172dd2
* v0.18.0 releasev0.18.0Sergey Shepelev2016-01-234-9/+16
|
* Mention all changes to be releasedJakub Stasiak2016-01-231-0/+62
| | | | I hope I didn't miss anything/anyone.
* travis: Fix allowing more CPU usageJakub Stasiak2016-01-231-0/+1
| | | | | | | | | The original commit[1] was not enough because we run tests using tox and tox by default doesn't pass all environment variables to the environments in which tests run - this patch whitelists subset of environment variables set by Travis, including one that we read and use. [1] 74f418bcf38ac128cadad641d59d7ecd8f127340
* greenio: Remove sendall-like semantincs from GreenSocket.sendJakub Stasiak2016-01-232-10/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When a socket timeout was set the sendall-like semantics was resulting in losing information about sent data and raising socket.timeout exception. The previous GreenSocket.send() implementation used to call fd.send() in a loop until all data was sent. The issue would manifest if at least one fd.send() call succeeded and was followed by a fd.send() call that would block and a trampoline that timed out. The client code would see socket.timeout being raised and would rightfully assume that no data was sent which would be incorrect. Discussed at https://github.com/eventlet/eventlet/issues/260. The original commit[1] has been reverted because it broke the test_multiple_readers test. After some debugging I believe I more or less know what's going on: The test uses sendall() which calls send() in a loop if send() reports only part of the data sent. Previously sendall() would call send() only one anyway because send() had sendall-like semantics; send has an internal loop on its own and, previously, it'd call the underlying socket object send() and, in case of partial write, call trampoline mechanism which would switch to another greenthread ready to run. After the change partial writes no longer result in trampoline mechanism being called which means that during this test it's likely that sendall() doesn't yield control to the hub even once. Similarly the current recv() implementation - it attemps to read from a socket first and only yields control to the hub if there's nothing to read at the moment; when one of the readers obtained control it's likely it'd manage to read all the data from the socket without yielding control to the hub and letting the other reader receive any data. The test changes the sending code so that it not only yields to the hub but also waits a bit so that both the readers have to yield to the hub when trying to read and there's no data available; the tests confirmed this lets both the readers receive some data from the socket which is the purpose of the test. [1] 4656eadfa5ae1237036a63ad4004dbee4572debf
* tests: Remove unused codeJakub Stasiak2016-01-231-5/+2
|
* Remove pip download cache referencesJakub Stasiak2016-01-232-3/+0
| | | | | Newest pip versions don't support it anymore and it's been deprecated for a while.
* Replaced broken link in design patterns exampleAzhar Hussain2016-01-131-1/+1
| | | | fixes https://github.com/eventlet/eventlet/issues/285
* patcher: Eliminate some redundancyJakub Stasiak2016-01-121-21/+13
| | | | | All those code blocks do the same thing so I decided to extract the logic into a one place.
* Fix whitespace to conform to our linting testsJakub Stasiak2016-01-122-2/+2
| | | | | | I completely missed those two. Fixes: 24d283cfdb3cb9e6f0b4cad3d676940b4a3ce252
* backdoor: Add Unix and IPv6 socket supportbackdoorEric Urban2016-01-112-6/+56
| | | | | BitBucket: https://github.com/eventlet/eventlet/issues/157 GitHub: https://github.com/eventlet/eventlet/pull/56
* green selectors: Override DefaultSelector as wellJakub Stasiak2016-01-112-0/+6
| | | | | | | | | | | | The original patch[1] missed one thing which I just discovered - there's selectors.DefaultSelector alias for the "the most efficient implementation available on the current platform"[2]. Before this patch a non-green selector class would be obtained when DefaultSelector was used. [1] 0d509ef7d2eea3ed4f8ade35d5d09918c811b56c [2] https://docs.python.org/3.5/library/selectors.html#selectors.DefaultSelector
* green select: Make sure devpoll method is removedJakub Stasiak2016-01-112-2/+2
| | | | | | | | devpoll is another polling method, added in Python 3.3. We don't have a green version of it so we better remove it too (see [1] for more details). [1] f63165c0e3c85699ebdb454878d1eaea13e90553
View Lorry Controller Status