Mailman 3 October 2001 - Twisted

[Twisted-Python] Netstring unit tests fail with Python 1.x
by Tommi Virtanen 06 Oct '01

06 Oct '01

Don't know if this is the right fix, but... [tv@ki ~/work/twistedmatrix/Twisted]$ DISPLAY= PYTHONPATH=. ./bin/runtests .....................E.E.................attached [twisted.test.test_pb.DummyPerspective] .............Rebuilding Delayed Event Queue... (rebuilding stale twisted.test.crash_test_dummy.X instance (do)) Rebuilt. ........Created New Participant: glyph Created New Participant: sean ..............................correct .... Time: 1.545s !!!FAILURES!!! Test Results Run: 94 ; Failures: 0; Errors: 2 There were 2 errors: 1) twisted.test.test_protocols.NetstringReceiverTestCase.testBuffer Traceback (innermost last): File "./twisted/test/test_protocols.py", line 127, in testBuffer a.dataReceived(s) File "./twisted/protocols/basic.py", line 89, in dataReceived self.doData() File "./twisted/protocols/basic.py", line 53, in doData buffer,self.__data = self.__data[:self.length],self.__data[self.length:] TypeError: slice index must be int 2) twisted.test.test_protocols.NetstringReceiverTestCase.testSafe Traceback (innermost last): File "./twisted/test/test_protocols.py", line 140, in testSafe r.dataReceived(s) File "./twisted/protocols/basic.py", line 122, in dataReceived NetstringReceiver.dataReceived(self, data) File "./twisted/protocols/basic.py", line 89, in dataReceived self.doData() File "./twisted/protocols/basic.py", line 53, in doData buffer,self.__data = self.__data[:self.length],self.__data[self.length:] TypeError: slice index must be int [tv@ki ~/work/twistedmatrix/Twisted]$ Index: basic.py =================================================================== RCS file: /cvs/Twisted/twisted/protocols/basic.py,v retrieving revision 1.10 diff -u -r1.10 basic.py --- basic.py 2001/09/30 18:55:43 1.10 +++ basic.py 2001/10/06 21:39:05 @@ -77,7 +77,7 @@ raise NetstringParseError self.__data = self.__data[m.end():] if m.group(1): - self.length = self.length * (10**len(m.group(1))) + long(m.group(1)) + self.length = self.length * (10**len(m.group(1))) + int(m.group(1)) if m.group(2): self.__buffer = '' self.mode = DATA [tv@ki ~/work/twistedmatrix/Twisted]$ DISPLAY= PYTHONPATH=. ./bin/runtests .......................................attached [twisted.test.test_pb.DummyPerspective] .............Rebuilding Delayed Event Queue... (rebuilding stale twisted.test.crash_test_dummy.X instance (do)) Rebuilt. ........Created New Participant: glyph Created New Participant: sean ..............................correct .... Time: 1.550s OK (94 tests) [tv@ki ~/work/twistedmatrix/Twisted]$ -- tv(a){{hq.yok.utu,havoc,gaeshido}.fi,{debian,wanderer}.org,stonesoft.com} double a,b=4,c;main(){for(;++a<2e6;c-=(b=-b)/a++);printf("%f\n",c);}

2 1

[Twisted-Python] Re: pretzel
by Kevin Turner 05 Oct '01

05 Oct '01

What's the "if" clause in the features loop of BreadStick.sendMessage() supposed to be doing? Why is the feature's method named "do" instead of "sendMessage"? What's the difference between features and SaltGrains? What's the difference between a SaltGrain (or feature) and a pb^H^Hpassport.Service? May saltGrain.network be the same as service.serviceName? May person be a passport.Identity? Must message be a string? See... Pretzel puzzles me. This project has some ideas with great potential, and it's precisely the sort of thing Twisted is good at... so I don't get why you seem to be intent on re-implementing Twisted inside of Twisted. I guess that would make Pretzel live up to its name, but... This is why I kept asking what features pretzel was going to have, because that's where I thought the fun was going to be. If you had an factoid-service, for example, which could be accessed through both IRC and the web, that could be cool. And there are definately some protocol/interface-specific gaps to fill in before that will work. i.e. msg'ing "Hi" to a Twisted IRC client doesn't currently set up any sort of broker/viewpoint for you, so that'd be an excellent area for �berbot code to cover... But establishing local services (SaltGrains) and passing messages between them... That's what Twisted *does*, no need to create more infastructure there. Am I missing something? I may just be speaking up too soon on a subject I know nothing about, but hey, that's what you get for checking it in to CVS. ;) "No, you're not going to trick me into writing IRC code *again*"-ly, - Kevin (Acapnotic) -- Kevin Turner <acapnotic(a)twistedmatrix.com> | OpenPGP encryption welcome The moon is 16.87 days old, waning gibbous, 95% illuminated.

4 5

[Twisted-Python] Can pipes be used in non-blocking mode?
by Matt Campbell 05 Oct '01

05 Oct '01

I noticed that twisted.internet.process currently uses pipes in blocking mode. In particular, when ProcessWriter calls os.write(), it blocks. Can writing be done in non-blocking mode instead? Thanks. -- Matt Campbell <http://www.pobox.com/~mattcampbell/>

2 1

[Twisted-Python] Interface documentation
by Itamar Shtull-Trauring 03 Oct '01

03 Oct '01

Right now, too much of the interface documentation is mixed up with the class implementation (if there *is* documentation). I propose making separate interface classes used only for documentation purposes. For example, in twisted.internet we might have interface.py with the following code, which would allow us to remove all the docstrings from abstract.FileDescriptor (that class is way too huge anyway), and maybe add a __implements__ = (interfaces.IProducer,) to the FileDescriptor class: twisted/internet/interfaces.py: ============================================================================== class IProducer: """A producer produces data for a consumer. If this is a streaming producer, it will only be asked to resume producing if it has been previously asked to pause. Also, if this is a streaming producer, it will ask the producer to pause when the buffer has reached a certain size. In other words, a streaming producer is expected to produce (write to this consumer) data in the main IO thread of some process as the result of a read operation, whereas a non-streaming producer is expected to produce data each time resumeProducing() is called. If this is a non-streaming producer, resumeProducing will be called immediately, to start the flow of data. Otherwise it is assumed that the producer starts out life unpaused. """ def resumeProducing(self): """Resume producing data. This tells a producer to re-add itself to the main loop and produce more data for its consumer. """ pass def pauseProducing(self): """Pause producing data. Tells a producer that it has produced too much data to process for the time being, and to stop until resumeProducing() is called. """ pass def stopProducing(self): """Stop producing data. This tells a producer that its consumer has died, so it must stop producing data for good. """ pass

1 1

[Twisted-Python] Twisted doesn't work in threaded mode using jython
by Itamar Shtull-Trauring 03 Oct '01

03 Oct '01

One issue is the lack of errno, which StringIO requires (solved by making an empty errno.py). So this can be solved trivially. A more serious issue is the lack of a _JavaWaker...

1 0

[Twisted-Python] Twisted T-Shirts
by Jason Asbahr 02 Oct '01

02 Oct '01

http://www.twisted.co.uk/orders/orderuk2_def.htm Ok, not quite, but after looking at this link, it's OBVIOUS we need to infiltrate IPC 10 wearing actual Twisted t-shifts. I favor black cotton. :-) Jason PS -- The Twisted Records designs are not a bad place to start. Simple, but interesting.

2 1

[Twisted-Python] New thread-based select replacement
by Itamar Shtull-Trauring 01 Oct '01

01 Oct '01

Great to see I can run on Java now - I'll test out the performance difference ASAP. Thanks glyph! I would like to point out that it might be nice to use this on CPython as well - both for performance comparision and for other platforms that don't support select() (BeOS? PlayStation 2? Amiga?). Haven't looked at the code yet so I don't know if that's dealt with.

1 0