Mailman 3 July 2010 - Python-Dev

stdlib socket usage and "keepalive"
by Jesus Cea July 12, 2010

July 12, 2010

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Debugging a strange problem today, I got the following result: Sockets open by stdlib libraries are open without the "keepalive" option, so the system default is used. The system default under linux is "no keepalive". So, if you are using a URLlib connection, POP3 connection, IMAP connection, etc., any stdlib that internally creates a socket, and your server goes away suddendly (you lose network connectivity, by instance), the library will wait FOREVER for the server. The client can't detect that the server is not longer available. The "keepalive" option will send a probe packed every X minutes of inactivity, to check if the other side is still alive, even if the connection is idle. The issue is bad, but the solution is simple enough. Options: 1. All "client" libraries should create sockets with the "KEEPALIVE" option. 2. Modify the socket C module to create all sockets as "Keepalive" by default. 3. To have a global variable in the socket module to change the default for future sockets. Something like current "socket.setdefaulttimeout()". The default should be "keepalive". 4. Modify client libraries to accept a new optional socket-like object as an optional parameter. This would allow things like transparent compression or encryption, or to replace the socket connection by anything else (read/write to shared memory or database, for example). This is an issue in Linux because by default the sockets are not "keepalive". In other Unix systems, the default is "keepalive". I don't know about MS Windows. What do you think?. The solution seems trivial, after deciding the right way to go. PS: "socket.setdefaulttimeout()" is not enough, because it could shutdown a perfectly functional connection, just because it was idle for too long. - -- Jesus Cea Avion _/_/ _/_/_/ _/_/_/ jcea(a)jcea.es - http://www.jcea.es/ _/_/ _/_/ _/_/ _/_/ _/_/ jabber / xmpp:jcea@jabber.org _/_/ _/_/ _/_/_/_/_/ . _/_/ _/_/ _/_/ _/_/ _/_/ "Things are not so easy" _/_/ _/_/ _/_/ _/_/ _/_/ _/_/ "My name is Dump, Core Dump" _/_/_/ _/_/_/ _/_/ _/_/ "El amor es poner tu felicidad en la felicidad de otro" - Leibniz -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQCVAwUBS8OV2Zlgi5GaxT1NAQIJhwP8CH+gij4KfrJ1oHW5ys4PboH5Ru0pgdly Wbsza0+uj3p68P1vDnC9jIr7j+fI3ql3DOc8zUoIKGpJoaWVspbv3c4vI4ATLHo+ J6I18dpkviRT8/sT/69vMvghaGndiO0Sks/S4tDjhNstYH7oGjWxi63cKqtGPY/p WSTLpwrd4SY= =vkT4 -----END PGP SIGNATURE-----

8 16

Can ftp url start with file:// ?
by Senthil Kumaran July 11, 2010

July 11, 2010

Strictly not a Python question, but I wanted to know from the experience of others in this list. Is this is valid ftp url? # file://ftp.example.com/blah.txt (an ftp URL) My answer is no. When we have the scheme specifically mentioned as file:// it is no point in considering it as ftp url (which should start with ftp://). If I go ahead with this assumption and fix a bug in stdlib, I am introducing a regression because at the moment the above is considered a ftp url. -- Senthil A real diplomat is one who can cut his neighbor's throat without having his neighbour notice it. -- Trygve Lie

7 14

I use both IDLE and my console
by Van July 11, 2010

July 11, 2010

Hello all, Buggy or not it's quite handy and as I said I use both when programming in python. To discontinue it is just not a good idea. Hausburn

1 0

Switching dev docs URL to Py3k?
by Nick Coghlan July 10, 2010

July 10, 2010

With the trunk closed to new development, should we switch http://docs.python.org/dev/ to show the docs built from the Py3k branch? (The in-development Py3k docs are currently at dev/py3k/) Cheers, Nick. -- Nick Coghlan | ncoghlan(a)gmail.com | Brisbane, Australia

3 3

Re: [Python-Dev] New regex module for 3.2?
by William Wahl July 9, 2010

July 9, 2010

H as long as we aren't the ones writing the check:) BJ ----------Original Message---------- From: Fred Drake <fdrake(a)acm.org> Sent: Fri, July 09, 2010 1:16 PM To: MRAB <python(a)mrabarnett.plus.com> Cc: Python-Dev <python-dev(a)python.org> Subject: Re: [Python-Dev] New regex module for 3.2? On Fri, Jul 9, 2010 at 3:35 PM, MRAB <python(a)mrabarnett.plus.com> wrote: > I concentrated my efforts on the matching speed because regexes tend to > be compiled only once, and are cached anyway, so I don't think it's as > important. I think most here will agree with that, but it might be good to keep in mind that the sre implementation has already made that trade-off once. I don't remember what the compilation slowdown was at the time, but I'd be surprised if Google can't find it, given sufficient fu. -Fred -- Fred L. Drake, Jr. <fdrake at gmail.com> "A storm broke loose in my mind." --Albert Einstein _______________________________________________ Python-Dev mailing list Python-Dev(a)python.org http://mail.python.org/mailman/listinfo/python-dev Unsubscribe: http://mail.python.org/mailman/options/python-dev/bj.wahl%40oracle.com

1 0

Summary of Python tracker Issues
by Python tracker July 9, 2010

July 9, 2010

ACTIVITY SUMMARY (2010-07-02 - 2010-07-09) Python tracker at http://bugs.python.org/ To view or respond to any of the issues listed below, click on the issue number. Do NOT respond to this message. 2813 open (+52) / 18217 closed (+15) / 21030 total (+67) Open issues with patches: 1133 Average duration of open issues: 706 days. Median duration of open issues: 503 days. Open Issues Breakdown open 2779 (+51) languishing 14 ( +0) pending 19 ( +1) Issues Created Or Reopened (75) _______________________________ backport 3.0-style \u/\U processing in raw strings when unicod 2010-07-07 http://bugs.python.org/issue2570 reopened alexandre.vassalotti patch audioop: check that length is a multiple of the size 2010-07-04 CLOSED http://bugs.python.org/issue7673 reopened pitrou patch sys.getwindowsversion as PyStructSequence 2010-07-08 CLOSED http://bugs.python.org/issue7766 reopened brian.curtin patch, patch, needs review dis module documentation gives no indication of the dangers of 2010-07-05 http://bugs.python.org/issue7829 reopened tjreedy needs review `make patchcheck` should check the whitespace of .c/.h files 2010-07-05 http://bugs.python.org/issue8912 reopened belopolsky patch, needs review Year range in timetuple 2010-07-05 http://bugs.python.org/issue9005 reopened belopolsky patch Separate compilation of time and datetime modules 2010-07-04 http://bugs.python.org/issue9012 reopened doko patch str.capitalize() should not lower the rest of the string 2010-07-04 CLOSED http://bugs.python.org/issue9091 reopened tjreedy patch test_coercion fails in refleak runs 2010-07-02 CLOSED http://bugs.python.org/issue9145 created pitrou Segfault in hashlib in OpenSSL FIPS mode using non-FIPS-compli 2010-07-02 http://bugs.python.org/issue9146 created dmalcolm patch dis.show_code() variant that accepts source strings (and retur 2010-07-03 http://bugs.python.org/issue9147 created ncoghlan os.execve puts process to background on windows 2010-07-03 http://bugs.python.org/issue9148 created techtonik colorsys.py function rgb_to_hls 2010-07-03 CLOSED http://bugs.python.org/issue9149 created David.Hood IDLE should not save trailing whitespace after strip trailing 2010-07-03 http://bugs.python.org/issue9150 created rhprogrammer Demo/classes/Dates.py does not work in 3.x 2010-07-03 CLOSED http://bugs.python.org/issue9151 created belopolsky patch, easy Dead code in datetime module 2010-07-04 CLOSED http://bugs.python.org/issue9152 created belopolsky patch, easy Run tests and demos as part of the test suite 2010-07-04 http://bugs.python.org/issue9153 created merwok Parser module doesn't understand function annotations. 2010-07-04 http://bugs.python.org/issue9154 created mark.dickinson patch Reserve COMPARE_OP for rich comparisons 2010-07-04 http://bugs.python.org/issue9155 created serprex patch socket._fileobject: read raises AttributeError when closed in 2010-07-05 http://bugs.python.org/issue9156 created panzi Allow inspection of used decorators on a function 2010-07-05 http://bugs.python.org/issue9157 created mitar PyArg_ParseTuple y* documentation is incorrect 2010-07-05 CLOSED http://bugs.python.org/issue9158 created terrence patch asyncore.file_wrapper does not register its fd to socket_map 2010-07-05 CLOSED http://bugs.python.org/issue9159 created kimhyunkang Documentation link of 3.2 not working at docs.python.org 2010-07-05 CLOSED http://bugs.python.org/issue9160 created eli.bendersky add_option in optparse no longer accepts unicode string 2010-07-05 http://bugs.python.org/issue9161 created cmcqueen1975 License for multiprocessing files 2010-07-05 http://bugs.python.org/issue9162 created FladischerMichael easy test_gdb fails 2010-07-05 CLOSED http://bugs.python.org/issue9163 created Michael.Blume 2.7 sysconfig should handle arch duplicates while building uni 2010-07-05 http://bugs.python.org/issue9164 created jyrkiwahlstedt Add math.isfinite() 2010-07-05 http://bugs.python.org/issue9165 created benjamin.peterson patch minor misstatement in What's New in 2.7 2010-07-05 CLOSED http://bugs.python.org/issue9166 created stutzbach patch argv double encoding on OSX 2010-07-05 http://bugs.python.org/issue9167 created piro patch setuid in smtp.py sheds privileges before binding port 2010-07-05 http://bugs.python.org/issue9168 created alfmel patch Deprecate use of more than 3 positional arguments in timedelta 2010-07-05 http://bugs.python.org/issue9169 created belopolsky zipfile cannot read AES encrypted files 2010-07-05 http://bugs.python.org/issue9170 created aymill sysconfig module does not support -m option 2010-07-05 http://bugs.python.org/issue9171 created giampaolo.rodola easy zipfile.extractall always raises an OSError after successfully 2010-07-05 http://bugs.python.org/issue9172 created aymill logger statement not guarded in shutil._make_tarball 2010-07-06 http://bugs.python.org/issue9173 created paulsmith patch unittest assertEqual error output sub-optimal for single strin 2010-07-06 http://bugs.python.org/issue9174 created r.david.murray patch ctypes doesn't build on hp-ux 2010-07-06 http://bugs.python.org/issue9175 created ronaldoussoren module termios doesn't build on HP-UX 2010-07-06 http://bugs.python.org/issue9176 created ronaldoussoren ssl.read/write on closed socket raises AttributeError 2010-07-06 http://bugs.python.org/issue9177 created cbay Compile problem on HP-UX 11 2010-07-06 CLOSED http://bugs.python.org/issue9178 created ronaldoussoren Lookback with group references incorrect (two issues?) 2010-07-06 http://bugs.python.org/issue9179 created acooke Memory Error 2010-07-06 http://bugs.python.org/issue9180 created freakcycle Solaris extension building does not work with 64 bit python 2010-07-06 http://bugs.python.org/issue9181 created redcomet document “--” as a way to distinguish option w/ narg='+' f 2010-07-06 http://bugs.python.org/issue9182 reopened merwok easy Intern UTC timezone 2010-07-06 http://bugs.python.org/issue9183 created belopolsky patch, easy open() doc: default value for buffering is -1, not None 2010-07-06 http://bugs.python.org/issue9184 created haypo patch os.getcwd causes infinite loop on solaris 2010-07-07 http://bugs.python.org/issue9185 created csernazs math.log1p(-1.0) produces OverflowError; should be ValueError 2010-07-07 CLOSED http://bugs.python.org/issue9186 created mark.dickinson patch Can't find file "vcvars64.bat" 2010-07-07 CLOSED http://bugs.python.org/issue9187 created Lloyd.Sheng test_gdb fails for UCS2 builds with UCS2 gdb 2010-07-07 http://bugs.python.org/issue9188 created pitrou patch Improve CFLAGS handling 2010-07-07 http://bugs.python.org/issue9189 created jyasskin Undefined behaviour in _PyFloat_Pack4 2010-07-07 http://bugs.python.org/issue9190 created mark.dickinson winreg.c:Reg2Py() may leak memory (in unusual circumstances) 2010-07-07 http://bugs.python.org/issue9191 created stutzbach Add a a way to determine float format 2010-07-07 http://bugs.python.org/issue9192 created benjamin.peterson Versioned .so files 2010-07-07 http://bugs.python.org/issue9193 created barry patch winreg:fixupMultiSZ should check that P < Q in the inner loop 2010-07-08 http://bugs.python.org/issue9194 created stutzbach Link in docs from "String Formatting Operations" to "Template 2010-07-08 http://bugs.python.org/issue9195 created cmcqueen1975 Improve docs for string interpolation "%s" re Unicode strings 2010-07-08 http://bugs.python.org/issue9196 created cmcqueen1975 subprocess module causing crash 2010-07-08 http://bugs.python.org/issue9197 created mpalluat Should repr() print unicode characters outside the BMP? 2010-07-08 http://bugs.python.org/issue9198 created amaury.forgeotdarc patch distutil upload command crashes when displaying server respons 2010-07-08 http://bugs.python.org/issue9199 created pje str.isprintable() is always False for large code points 2010-07-08 http://bugs.python.org/issue9200 created amaury.forgeotdarc IDLE: raises Exception TclError in a special case 2010-07-08 http://bugs.python.org/issue9201 created doko Update platform.win32_ver() to account for change to #8413 2010-07-08 CLOSED http://bugs.python.org/issue9202 created brian.curtin patch, needs review Use computed gotos by default 2010-07-08 http://bugs.python.org/issue9203 created pitrou The documentation of PyType_Type in py3k mentions types.TypeTy 2010-07-08 http://bugs.python.org/issue9204 created durban Parent process hanging in multiprocessing if children terminat 2010-07-08 http://bugs.python.org/issue9205 created gdb patch repr(os.stat(..)) is broken by the structseq change 2010-07-08 CLOSED http://bugs.python.org/issue9206 created belopolsky patch multiprocessing occasionally spits out exception during shutdo 2010-07-08 http://bugs.python.org/issue9207 created gdb SMTPHandler in the logging module does not handle unicode stri 2010-07-08 http://bugs.python.org/issue9208 created norbidur pstats module crashes on trailing backslash 2010-07-09 http://bugs.python.org/issue9209 created isandler remove --with-wctype-functions configure option 2010-07-09 http://bugs.python.org/issue9210 created amaury.forgeotdarc patch Cannot close socket after ssl handshake failed 2010-07-09 CLOSED http://bugs.python.org/issue9211 created metachris Issues Now Closed (59) ______________________ Broken pipes should be handled better in 2.x 940 days http://bugs.python.org/issue1596 tjreedy Backport of PEP 3102 "keyword-only arguments" to 2.6 912 days http://bugs.python.org/issue1745 alexandre.vassalotti patch email.MIMEText.MIMEText.as_string incorrectly folding long sub 888 days http://bugs.python.org/issue1974 nicdumz test_al converted to unittest 884 days http://bugs.python.org/issue2022 benjamin.peterson patch test_cd converted to unittest 884 days http://bugs.python.org/issue2023 benjamin.peterson patch test_gl.py converted to unittest 884 days http://bugs.python.org/issue2024 benjamin.peterson patch test_sunaudiodev.py converted to unittest 883 days http://bugs.python.org/issue2044 benjamin.peterson patch Python should expose a pthread_cond_timedwait API for threadin 748 days http://bugs.python.org/issue3155 pitrou Cleanup of tp_basicsize inheritance 732 days http://bugs.python.org/issue3268 benjamin.peterson patch Fork + shelve causes shelve corruption and backtrace 566 days http://bugs.python.org/issue4679 belopolsky Error in atexit._run_exitfuncs [...] Exception expected for va 527 days http://bugs.python.org/issue5089 mark.dickinson urlencode does not handle "bytes" and could easily handle alte 480 days http://bugs.python.org/issue5468 orsenthil patch Implement a way to change the python process name 458 days http://bugs.python.org/issue5672 piro patch What should happen to signals when the main thread dies? 402 days http://bugs.python.org/issue6162 tjreedy Enhance dis.dis to autocompile codestrings 357 days http://bugs.python.org/issue6507 merwok patch, easy Do we have something like os.pid_exists()? 350 days http://bugs.python.org/issue6554 tjreedy interrupts during long writes cause connection corruption with 346 days http://bugs.python.org/issue6587 pitrou otparse: processing variable number of short option arguments 346 days http://bugs.python.org/issue6592 tjreedy audioop: check that length is a multiple of the size 0 days http://bugs.python.org/issue7673 mark.dickinson patch sys.getwindowsversion as PyStructSequence 0 days http://bugs.python.org/issue7766 brian.curtin patch, patch, needs review Fnmatch cache is never cleared during usage 156 days http://bugs.python.org/issue7846 r.david.murray patch, needs review MemoryError While Executing Python Code 146 days http://bugs.python.org/issue7899 mark.dickinson String interpolation doesn't work with sys.version_info 84 days http://bugs.python.org/issue8413 belopolsky patch Let cmd.cmd.intro be unicode friendly 76 days http://bugs.python.org/issue8470 r.david.murray itertools.filterfalse() function missing 75 days http://bugs.python.org/issue8472 georg.brandl test_gdb, gdb/libpython.py: Unable to read information on pyth 78 days http://bugs.python.org/issue8482 dmalcolm test_gdb can fail with compiler opts 66 days http://bugs.python.org/issue8605 pitrou patch 2to3 fails with assertion failure on "from itertools import *" 31 days http://bugs.python.org/issue8892 benjamin.peterson patch PyBuffer_Release signature in 3.1 documentation is incorrect 30 days http://bugs.python.org/issue8946 krisvale Provide parseable repr to datetime.timezone 21 days http://bugs.python.org/issue9000 belopolsky patch, easy 2.7: eval hangs on AIX 16 days http://bugs.python.org/issue9020 pitrou patch At least some Tools utilities are still Python 2 9 days http://bugs.python.org/issue9083 merwok str.capitalize() should not lower the rest of the string 1 days http://bugs.python.org/issue9091 orsenthil patch Make python-m pickletools do something useful 9 days http://bugs.python.org/issue9094 belopolsky patch, easy importlib doesn't handle valid marshalled data with invalid se 7 days http://bugs.python.org/issue9096 brett.cannon remove numbers from 3-.. level entries in docs toc 5 days http://bugs.python.org/issue9106 tjreedy Parser module fails to validate class decorators correctly. 4 days http://bugs.python.org/issue9128 mark.dickinson patch Parser module fails on short relative imports 4 days http://bugs.python.org/issue9130 mark.dickinson patch RuntimeError when profiling Decimal 8 days http://bugs.python.org/issue9136 mark.dickinson patch Add examples for str.format() 1 days http://bugs.python.org/issue9139 ezio.melotti patch Import error for multiprocessing in 2.7rc2 on Windows 1 days http://bugs.python.org/issue9144 jnoller patch test_coercion fails in refleak runs 2 days http://bugs.python.org/issue9145 flox colorsys.py function rgb_to_hls 2 days http://bugs.python.org/issue9149 David.Hood Demo/classes/Dates.py does not work in 3.x 1 days http://bugs.python.org/issue9151 merwok patch, easy Dead code in datetime module 1 days http://bugs.python.org/issue9152 belopolsky patch, easy PyArg_ParseTuple y* documentation is incorrect 1 days http://bugs.python.org/issue9158 terrence patch asyncore.file_wrapper does not register its fd to socket_map 0 days http://bugs.python.org/issue9159 kimhyunkang Documentation link of 3.2 not working at docs.python.org 0 days http://bugs.python.org/issue9160 georg.brandl test_gdb fails 3 days http://bugs.python.org/issue9163 dmalcolm minor misstatement in What's New in 2.7 0 days http://bugs.python.org/issue9166 georg.brandl patch Compile problem on HP-UX 11 0 days http://bugs.python.org/issue9178 loewis math.log1p(-1.0) produces OverflowError; should be ValueError 0 days http://bugs.python.org/issue9186 mark.dickinson patch Can't find file "vcvars64.bat" 0 days http://bugs.python.org/issue9187 skrah Update platform.win32_ver() to account for change to #8413 0 days http://bugs.python.org/issue9202 brian.curtin patch, needs review repr(os.stat(..)) is broken by the structseq change 0 days http://bugs.python.org/issue9206 benjamin.peterson patch Cannot close socket after ssl handshake failed 0 days http://bugs.python.org/issue9211 metachris logging handlers raise exception on level 2367 days http://bugs.python.org/issue876421 Mickey.Killianey socket_htons does not work under AIX 64-bit 2299 days http://bugs.python.org/issue921868 skrah easy Add an updating load function in pickle 2230 days http://bugs.python.org/issue960821 belopolsky Top Issues Most Discussed (10) ______________________________ 20 Regexp 2.7 (modifications to current re 2.2.2) 815 days open http://bugs.python.org/issue2636 19 Use Py_UCS4 instead of Py_UNICODE in unicodectype.c 523 days open http://bugs.python.org/issue5127 18 Add pure Python implementation of datetime module to CPython 137 days open http://bugs.python.org/issue7989 16 multiprocessing occasionally spits out exception during shutdow 1 days open http://bugs.python.org/issue9207 13 Reserve COMPARE_OP for rich comparisons 5 days open http://bugs.python.org/issue9155 11 Should repr() print unicode characters outside the BMP? 1 days open http://bugs.python.org/issue9198 11 String interpolation doesn't work with sys.version_info 84 days closed http://bugs.python.org/issue8413 10 Fnmatch cache is never cleared during usage 156 days closed http://bugs.python.org/issue7846 9 difflib.SequenceMatcher not matching long sequences 773 days open http://bugs.python.org/issue2986 8 2.7 sysconfig should handle arch duplicates while building univ 4 days open http://bugs.python.org/issue9164

1 0

Py_Buffer and pep3118
by Kristján Valur Jónsson July 9, 2010

July 9, 2010

Hello there. I´m in the process of upgrading code to use python 2.7 and the new Py_Buffer objects. I have, however, come across an inconsistency in the documentation, both in the pep and the python docs, with the actual implementation. In addition to the different function signatures mentioned in issue 8946, the Py_Buffer member 'obj' is undocumented. The inline comment in object.h mentiones that it is a "borrowed reference" but this appears inconsistent with the implementation of PyBuffer_Release(). In short, what the documentation fails to mention (and the pep) is whether posessing a locked Py_Buffer structure also constitutes holding a reference to the exporting object? I think it does, but is this guaranteed to be so? Also, is the PyBuffer_Release() guaranteed to be "safe" to call multiple times (as the current implementation is)? Kristján

2 2

Call for Applications - PSF Sponsored Sprints
by Jesse Noller July 8, 2010

July 8, 2010

[Sending this to Python-Dev, as it might very well be of interest as the sprints are focused on Python "core" tasks] The PSF is happy to open our first call for applications for sprint funding! Have you ever had a group of people together to hack towards a common goal? You've hosted a sprint! Have you ever wanted to get a group of like minded Pythonistas together to hack for a day? You're going to want to hold a sprint! Whether you call them Sprints, Hackfests, Hack-a-thons, or any other name, they're a great way to hang out with like-minded developers and work on common code. Sprints are an unbeatable way to build friendships and contacts that will last for years to come, and they're a great way to learn about something new if you're just starting out. The Python Software Foundation has set aside funds to be distributed to world-wide sprint efforts. We're anticipating 2-3 events per month focused on covering topics to help the entire community: - Python Core bug triage and patch submission (on-boarding new contributors) - Python Core documentation (including process documentation) improvements - Porting libraries/applications to Python 3 - Python website/wiki content improvements - PyPI packaging hosting site improvements - Contribution to other "core" projects, such as packaging related issues. If you are interested in holding a sprint on any of the topics above and you're looking for some money to help out with sprint costs, we can help (up to a max of $250 USD). Prepare an application including the following information: - Date and Location: Where will the event be? What day and time? - Organizers: Who are the event organizers and sprint coach? Is the sprint being run by a Python user group? - Attendees: How many participants do you expect? - Goal: What is the focus and goal of the sprint? - Budget: How much funding you are requesting, and what will you use it for? - Applications should be sent to: sprints(a)python.org with the subject "Sprint Funding Application - <location>" We encourage anyone - even those who have never held, or been to a sprint - to consider holding one. We will help you as much as we can with welcome packets, advertising, and hooking you up with required resources - anything to make it possible. As part of being approved, the you will need to agree to deliver a report (hopefully, with pictures!) of the sprint to the Sprint Committee, so we can post it on the sprint blog and site: http://www.pythonsprints.com If you have any questions or need more information, contact us by email at sprints(a)python.org. More information is up on our blog: http://pythonsprints.com/2010/07/8/call-applications-now-open/

1 0

Licensing // PSF // Motion of non-confidence
by anatoly techtonik July 7, 2010

July 7, 2010

Sorry for touching a sore point of if I sound like a boss to someone. I tried to be as constructive as possible, but politeness was not the point, so I can only hope you understand. I do not think PSF does its job well and here is why. 1. Python licensing terms are explained poorly In order to "advance the Python programming language", and "to support and facilitate the growth of international community" PSF should be clear about the legal stuff with no complications. So that even people without English background can understand the terms. 1.1 Let's start with http://www.python.org/psf/license/ ... Full text of the Python 2.6.2 license ... Q: Why 2.6.2? A: Nobody cares. Inside referenced doc there are several licenses for Python + historical note. I remember I've tried to read this file several times, but only after comparing freeness and beerness of PSF license to GPL in private talk with Eric I was able to go through this seriously. Moving on... 1.2. Mixed content of Python License file Attached or see http://code.python.org/hg/branches/release2.7-maint/annotate/41a3e598052f/L… The file consists of several licenses for multiple versions of Python. It is an unusual mix that negatively affects understanding. For the sake of clarity the LICENSE file scope should only be relevant to the current code. It may include a historical reference to other files if needed - LICENSE-HISTORY if you like. It is important to keep LICENSE terms concise for occasional contributors with skills needed for core development, because for outsiders there are many open source projects out there. _For example_, if I see an interesting project with GPL license (Ruby) and BSD license (Go) and ask myself "What should I choose for the next time I feel too depressive to spend my weekend watching movies?" - I will choose Go. I am sure it won't be a waste of my life time, just because both languages are equally trendy (Ruby sites are truly Gems in design, and Go presentation at Google IO was awesome), so license here is deciding factor. Every language has bugs and it is inevitable that you would like them to be fixed. As it is open source.. yes - you'll be politely asked to go fix them yourself. If you're an expert - you will fix them anyway (whatever you share fixes or not is a different story). Over the time you'll become acquainted with codebase and of course you will want to use this experience in your daily job writing code for customers who pay you money for keeping the code closed. In this case GPL allows you the freedom to rewrite the code from scratch, BSD - to copy/paste with attribution notice. If I'd be an outsider - when I'd see PSF license - I'd think - "or no, thats yet another stupid big license" (YASBL) and move on to the next news item (this time from pure C universe). 1.3 Python License content I believe that contents of attached LICENSE-PYTHON-2.x single copy/paste from LICENSE is enough for understanding and complying for Python 2.1+ series. Correct me if I wrong. But thanks to Berne Convention the wording can and should be improved. I hope the example in 1.2 made it clear that to attract professionals with strong core development skills, who are able to make quality contributions and are usually sought after in commercial projects, it is important to have license terms similar to those people already familiar and comfortable with, i.e. BSD, MIT/X (or more recent ISC variant - http://en.wikipedia.org/wiki/ISC_license) I propose the simplified PSF License v3 that in comparison to PSF v2: 1.3.1 combines point 4 and 5 to standard "THE SOFTWARE IS PROVIDED AS-IS ..." copied verbatim from ISC (MIT equivalent) license 1.3.2 drops point 6 as it lacks the copyright of Captain Obvious 1.3.3 combines points 8, 1 and 2 into standard ISC (simplified MIT) header with "and the following conditions are met:" from BSD license to outline the requirements of PSF in comparison to MIT, BSD and ISC style licenses 1.3.4 replaces point 7 with standard "Neither the name of the <organization>... " from BSD license 1.3.5 replaces point 3 with "Redistributions of derivative works should include a brief summary of the changes made to Python" Full text of license is attached as PSF-LICENSE-v3. For those who complain that I am not sending patches and only direct people what to do - LICENSE.patch contains the patch made with diff.py tool that comes bundled with Python. Please indicate if and where the license became different from legal point of view. 2. PSF fails to comply with the terms of Contributors Agreement http://python.org/psf/contrib/contrib-form/ Note that there are also: http://python.org/psf/contrib/contrib-form-python/ http://python.org/psf/contrib/contrib-form-jython/ which are reachable by "python contributor's agreement" search query and it is doesn't clear why there are multiple terms and why they differ. -- personal intermezzo -- The fact that Contributors Agreements chapter is here is that I didn't care about the license much until people sorted me out from Python community of "those who signed contributor's agreement" into "those other outcasts that don't have the right to speak" category. I started to ask questions about licensing stuff, and received a bunch of negative replies like "why don't you sign the agreement first?", "we've consulted with lawyer - believe us", "if you are not a lawyer - why are you asking?". I do not know how many of these replies were from members of PSF, but they surely were not happy to clarify anything. I stopped sending patches (because I didn't sign this agreement) and was accused in unethical behavior like doing myself nothing and just directing people what to do. To prevent further allegations I hereby claim that PSF is fully responsible in the outcome of current situation and in the people pissed off from Python Community as a result! Just kidding - a little but tired already. -- personal intermezzo -- I assume that pydotorg contributions, Python contributions and Jython are the same contributions, so there is only one Contributors Agreement, but there is no proof. PSF doesn't comply with the terms of agreement, because it requires contributors to submit code under either Apache 2.0 or AFL license. Both these licenses require to include text of license in distributions, but so far no Apache or AFL license text is present in Python source code tree. PSF also fails to explain they need to receive contributions using two very verbose licenses, which are almost identical. AFL in fact was found "redundant" in favor of Apache 2.0 in 2006 by OSI's License Proliferation Committee. PSF can just say: """Contributor understands and agrees that PSF shall have the irrevocable and perpetual right to make and distribute copies of any Contribution, as well as to create and distribute collective works and derivative works of any Contribution.""" That is enough. Why playing strange Apache 2.0 and AFL games if it is clearly written : """ Contributor understands and agrees that PSF shall have ... the right ... to create ... derivative works ... under the Initial License or under any other open source license approved by a unanimous vote of the PSF board """ " Contributor understands and agrees ... " - I do not understand, so I can't sign. Even if I sign CA there is a chance that I (or whoever has more money) can prove that terms impossible to understand. In this case all contributors agreements will be deemed invalid. So Python code is in danger and hence my derived work is in danger too. That's why I asked who supplied PSF lawyer to create such agreement. The argumentation that this agreement is legitimate like "Believe us - we've payed to lawyer" is surely not the thing I expected to hear. If you want international recognition of legitimacy of such agreements. - you need the reputation of some legal authority like Ernst&Young or Deloitte or one of big PSF supporters. In this case PSF will be eligible for trust instead of non-confidence notes. There are other reasons that make me hesitate about signing PSF agreements. For instance PSF doesn't provide any guarantees that: 1. The text of Contributors Agreement won't be changed after I sign it 2. My personal details are properly guarded and won't be shared with any members of PSF committee or partner companies or any third parties without my explicit permission That's all. You may want to reread the first paragraph of the letter before reply. P.S. Sorry for not replying to recent threads (i.e. Mercurial) - writing a letter like these takes from 4 to 12 hours and is very exhausting. -- anatoly t.

21 33

thoughts on the bytes/string discussion
by Bill Janssen July 7, 2010

July 7, 2010

Here are a couple of ideas I'm taking away from the bytes/string discussion. First, it would probably be a good idea to have a String ABC. Secondly, maybe the string situation in 2.x wasn't as broken as we thought it was. In particular, those who deal with lots of encoded strings seemed to find it handy, and miss it in 3.x. Perhaps strings are more like numbers than we think. We have separate types for int, float, Decimal, etc. But they're all numbers, and they all cross-operate. In 2.x, it seems there were two missing features: no encoding attribute on str, which should have been there and should have been required, and the default encoding being "ASCII" (I can't tell you how many times I've had to fix that issue when a non-ASCII encoded str was passed to some output function). So maybe having a second string type in 3.x that consists of an encoded sequence of bytes plus the encoding, call it "estr", wouldn't have been a bad idea. It would probably have made sense to have estr cooperate with the str type, in the same way that two different kinds of numbers cooperate, "promoting" the result of an operation only when necessary. This would automatically achieve the kind of polymorphic functionality that Guido is suggesting, but without losing the ability to do x = e(ASCII)"bar" a = ''.join("foo", x) (or whatever the syntax for such an encoded string literal would be -- I'm not claiming this is a good one) which presume would bind "a" to a Unicode string "foobar" -- have to work out what gets promoted to what. The language moratorium kind of makes this all theoretical, but building a String ABC still would be a good start, and presumably isn't forbidden by the moratorium. Bill

19 38