Mailman 3 March 2016 - Python-Dev

PEP 484: updates to Python 2.7 signature syntax
by Guido van Rossum March 21, 2016

March 21, 2016

PEP 484 was already updated to support signatures as type comments in Python 2.7. I'd like to add two more variations to this spec, both of which have already come up through users. First, https://github.com/python/typing/issues/188. This augments the format of signature type comments to allow (...) instead of an argument list. This is useful to avoid having to write (Any, Any, Any, ..., Any) for a long argument list if you don't care about the argument types but do want to specify the return type. It's already implemented by mypy (and presumably by PyCharm). Example: def gcd(a, b): # type: (...) -> int <code here> Second, https://github.com/python/typing/issues/186. This builds on the previous syntax but deals with the other annoyance of long argument lists, this time in case you *do* care about the types. The proposal is to allow writing the arguments one per line with a type comment on each line. This has been implemented in PyCharm but not yet in mypy. Example: def gcd( a, # type: int b, # type: int ): # type: (...) -> int <code here> In both cases we've considered a few alternatives and ended up agreeing on the best course forward. If you have questions or feedback on either proposal it's probably best to just add a comment to the GitHub tracker issues. A clarification of the status of PEP 484: it was provisionally accepted in May 2015. Having spent close to a year pondering it, and the last several months actively using it at Dropbox, I'm now ready to move with some improvements based on these experiences (and those of others who have started to use it). We already added the basic Python 2.7 compatible syntax (see thread starting at https://mail.python.org/pipermail/python-ideas/2016-January/037704.html), and having used that for a few months the two proposals mentioned above handle a few corner cases that were possible but a bit awkward in our experience. -- --Guido van Rossum (python.org/~guido)

2 3

PEP 484: a "NewType" constructor
by Guido van Rossum March 21, 2016

March 21, 2016

Here's one more thing we'd like to add to PEP 484. The description is best gleaned from the issue, in particular https://github.com/python/mypy/issues/1284#issuecomment-199021176 and following (we're going with option (A)). Really brief example: from typing import NewType UserId = NewType('UserId', int) Now to the type checker UserId is a new type that's compatible with int, but converting an int to a UserId requires a special cast form, UserId(x). At runtime UserId instances are just ints (not a subclass!) and UserId() is a dummy function that just returns its argument. For use cases see the issue. Also send feedback there please. -- --Guido van Rossum (python.org/~guido)

1 1

bitfields - short - and xlc compiler
by Michael Felt March 20, 2016

March 20, 2016

a) hope this is not something you expect to be on -list, if so - my apologies! Getting this message (here using c99 as compiler name, but same issue with xlc as compiler name) c99 -qarch=pwr4 -qbitfields=signed -DNDEBUG -O -I. -IInclude -I./Include -I/data/prj/aixtools/python/python-2.7.11.2/Include -I/data/prj/aixtools/python/python-2.7.11.2 -c /data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c -o build/temp.aix-5.3-2.7/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.o "/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c", line 387.5: 1506-009 (S) Bit field M must be of type signed int, unsigned int or int. "/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c", line 387.5: 1506-009 (S) Bit field N must be of type signed int, unsigned int or int. "/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c", line 387.5: 1506-009 (S) Bit field O must be of type signed int, unsigned int or int. "/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c", line 387.5: 1506-009 (S) Bit field P must be of type signed int, unsigned int or int. "/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c", line 387.5: 1506-009 (S) Bit field Q must be of type signed int, unsigned int or int. "/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c", line 387.5: 1506-009 (S) Bit field R must be of type signed int, unsigned int or int. "/data/prj/aixtools/python/python-2.7.11.2/Modules/_ctypes/_ctypes_test.c", line 387.5: 1506-009 (S) Bit field S must be of type signed int, unsigned int or int. for: struct BITS { int A: 1, B:2, C:3, D:4, E: 5, F: 6, G: 7, H: 8, I: 9; short M: 1, N: 2, O: 3, P: 4, Q: 5, R: 6, S: 7; }; in short xlC v11 does not like short (xlC v7 might have accepted it, but "32-bit machines were common then". I am guessing that 16-bit is not well liked on 64-bit hw now. reference for xlC v7, where short was (apparently) still accepted: http://www.serc.iisc.ernet.in/facilities/ComputingFacilities/systems/cluste… I am taking this is from xlC v7 documentation from the URL, not because I know it personally. So - my question: if "short" is unacceptable for POWER, or maybe only xlC (not tried with gcc) - how terrible is this, and is it possible to adjust the test so - the test is accurate? I am going to modify the test code so it is struct BITS { signed int A: 1, B:2, C:3, D:4, E: 5, F: 6, G: 7, H: 8, I: 9; unsigned int M: 1, N: 2, O: 3, P: 4, Q: 5, R: 6, S: 7; }; And see what happens - BUT - what does this have for impact on python - assuming that "short" bitfields are not supported? p.s. not submitting this a bug (now) as it may just be that "you" consider it a bug in xlC to not support (signed) short bit fields. p.p.s. Note: xlc, by default, considers bitfields to be unsigned. I was trying to force them to signed with -qbitfields=signed - and I still got messages. So, going back to defaults.

4 6

What does a double coding cookie mean?
by Guido van Rossum March 20, 2016

March 20, 2016

I came across a file that had two different coding cookies -- one on the first line and one on the second. CPython uses the first, but mypy happens to use the second. I couldn't find anything in the spec or docs ruling out the second interpretation. Does anyone have a suggestion (apart from following CPython)? Reference: https://github.com/python/mypy/issues/1281 -- --Guido van Rossum (python.org/~guido)

14 41

PEP 484 update: add Type[T]
by Guido van Rossum March 20, 2016

March 20, 2016

There's a more fundamental PEP 484 update that I'd like to add. The discussion is in https://github.com/python/typing/issues/107. Currently we don't have a way to talk about arguments and variables whose type is itself a type or class. The only annotation you can use for this is 'type' which says "this argument/variable is a type object" (or a class). But it's often useful to be able to say "this is a class and it must be a subclass of X". In fact this was proposed in the original rounds of discussion about PEP 484, but at the time it felt too far removed from practice to know quite how it should be used, so I just put it off. But it's been one of the features that's been requested most by the early adopters of PEP 484 at Dropbox. So I'd like to add it now. At runtime this shouldn't do much; Type would be just a generic class of one parameter that records its one type parameter. The real magic would happen in the type checker, which will be able to use types involving Type. It should also be possible to use this with type variables, so we could write e.g. T = TypeVar('T', bound=int) def factory(c: Type[T]) -> T: <implementation> This would define factory() as a function whose argument must be a subclass of int and returning an instance of that subclass. (The bound= option to TypeVar() is already described in PEP 484, although mypy hasn't implemented it yet.) (If I screwed up this example, hopefully Jukka will correct me. :-) Again, I'd like this to go out with 3.5.2, because it requires adding something to typing.py (and again, that's allowed because PEP 484 is provisional -- see PEP 411 for an explanation). -- --Guido van Rossum (python.org/~guido)

1 0

Summary of Python tracker Issues
by Python tracker March 18, 2016

March 18, 2016

ACTIVITY SUMMARY (2016-03-11 - 2016-03-18) Python tracker at http://bugs.python.org/ To view or respond to any of the issues listed below, click on the issue. Do NOT respond to this message. Issues counts and deltas: open 5459 ( +5) closed 32885 (+43) total 38344 (+48) Open issues with patches: 2375 Issues opened (36) ================== #15660: Clarify 0 prefix for width specifier in str.format doc, http://bugs.python.org/issue15660 reopened by terry.reedy #22758: Regression in Python 3.2 cookie parsing http://bugs.python.org/issue22758 reopened by berker.peksag #25934: ICC compiler: ICC treats denormal floating point numbers as 0. http://bugs.python.org/issue25934 reopened by zach.ware #26270: Support for read()/write()/select() on asyncio http://bugs.python.org/issue26270 reopened by gvanrossum #26481: unittest discovery process not working without .py source file http://bugs.python.org/issue26481 reopened by rbcollins #26541: Add stop_after parameter to setup() http://bugs.python.org/issue26541 opened by memeplex #26543: imaplib noop Debug http://bugs.python.org/issue26543 opened by Stephen.Evans #26544: platform.libc_ver() returns incorrect version number http://bugs.python.org/issue26544 opened by Thomas.Waldmann #26545: os.walk is limited by python's recursion limit http://bugs.python.org/issue26545 opened by Thomas.Waldmann #26546: Provide translated french translation on docs.python.org http://bugs.python.org/issue26546 opened by sizeof #26547: Undocumented use of the term dictproxy in vars() documentation http://bugs.python.org/issue26547 opened by sizeof #26549: co_stacksize is calculated from unoptimized code http://bugs.python.org/issue26549 opened by ztane #26550: documentation minor issue : "Step back: WSGI" section from "HO http://bugs.python.org/issue26550 opened by Alejandro Soini #26552: Failing ensure_future still creates a Task http://bugs.python.org/issue26552 opened by gordon #26553: Write HTTP in uppercase http://bugs.python.org/issue26553 opened by Sudheer Satyanarayana #26554: PC\bdist_wininst\install.c: Missing call to fclose() http://bugs.python.org/issue26554 opened by maddin200 #26556: Update expat to 2.2.1 http://bugs.python.org/issue26556 opened by christian.heimes #26557: dictviews methods not present on shelve objects http://bugs.python.org/issue26557 opened by Michael Crouch #26559: logging.handlers.MemoryHandler flushes on shutdown but not rem http://bugs.python.org/issue26559 opened by David Escott #26560: Error in assertion in wsgiref.handlers.BaseHandler.start_respo http://bugs.python.org/issue26560 opened by inglesp #26565: [ctypes] Add value attribute to non basic pointers. http://bugs.python.org/issue26565 opened by memeplex #26566: Failures on FreeBSD CURRENT buildbot http://bugs.python.org/issue26566 opened by haypo #26567: ResourceWarning: Use tracemalloc to display the traceback wher http://bugs.python.org/issue26567 opened by haypo #26568: Add a new warnings.showmsg() function taking a warnings.Warnin http://bugs.python.org/issue26568 opened by haypo #26571: turtle regression in 3.5 http://bugs.python.org/issue26571 opened by Ellison Marks #26574: replace_interleave can be optimized for single character byte http://bugs.python.org/issue26574 opened by Josh Snider #26576: Tweak wording of decorator docos http://bugs.python.org/issue26576 opened by Rosuav #26577: inspect.getclosurevars returns incorrect variable when using c http://bugs.python.org/issue26577 opened by Ryan Fox #26578: Bad BaseHTTPRequestHandler response when using HTTP/0.9 http://bugs.python.org/issue26578 opened by xiang.zhang #26579: Support pickling slots in subclasses of common classes http://bugs.python.org/issue26579 opened by serhiy.storchaka #26581: Double coding cookie http://bugs.python.org/issue26581 opened by serhiy.storchaka #26582: asyncio documentation links to wrong CancelledError http://bugs.python.org/issue26582 opened by awilfox #26584: pyclbr module needs to be more flexible on loader support http://bugs.python.org/issue26584 opened by eric.snow #26585: Use html.escape to replace _quote_html in http.server http://bugs.python.org/issue26585 opened by xiang.zhang #26586: Simple enhancement to BaseHTTPRequestHandler http://bugs.python.org/issue26586 opened by xiang.zhang #26587: Possible duplicate entries in sys.path if .pth files are used http://bugs.python.org/issue26587 opened by tds333 Most recent 15 issues with no replies (15) ========================================== #26584: pyclbr module needs to be more flexible on loader support http://bugs.python.org/issue26584 #26582: asyncio documentation links to wrong CancelledError http://bugs.python.org/issue26582 #26581: Double coding cookie http://bugs.python.org/issue26581 #26579: Support pickling slots in subclasses of common classes http://bugs.python.org/issue26579 #26577: inspect.getclosurevars returns incorrect variable when using c http://bugs.python.org/issue26577 #26571: turtle regression in 3.5 http://bugs.python.org/issue26571 #26565: [ctypes] Add value attribute to non basic pointers. http://bugs.python.org/issue26565 #26559: logging.handlers.MemoryHandler flushes on shutdown but not rem http://bugs.python.org/issue26559 #26556: Update expat to 2.2.1 http://bugs.python.org/issue26556 #26554: PC\bdist_wininst\install.c: Missing call to fclose() http://bugs.python.org/issue26554 #26550: documentation minor issue : "Step back: WSGI" section from "HO http://bugs.python.org/issue26550 #26546: Provide translated french translation on docs.python.org http://bugs.python.org/issue26546 #26543: imaplib noop Debug http://bugs.python.org/issue26543 #26541: Add stop_after parameter to setup() http://bugs.python.org/issue26541 #26539: frozen executables should have an empty path http://bugs.python.org/issue26539 Most recent 15 issues waiting for review (15) ============================================= #26586: Simple enhancement to BaseHTTPRequestHandler http://bugs.python.org/issue26586 #26585: Use html.escape to replace _quote_html in http.server http://bugs.python.org/issue26585 #26581: Double coding cookie http://bugs.python.org/issue26581 #26579: Support pickling slots in subclasses of common classes http://bugs.python.org/issue26579 #26576: Tweak wording of decorator docos http://bugs.python.org/issue26576 #26574: replace_interleave can be optimized for single character byte http://bugs.python.org/issue26574 #26568: Add a new warnings.showmsg() function taking a warnings.Warnin http://bugs.python.org/issue26568 #26567: ResourceWarning: Use tracemalloc to display the traceback wher http://bugs.python.org/issue26567 #26560: Error in assertion in wsgiref.handlers.BaseHandler.start_respo http://bugs.python.org/issue26560 #26553: Write HTTP in uppercase http://bugs.python.org/issue26553 #26547: Undocumented use of the term dictproxy in vars() documentation http://bugs.python.org/issue26547 #26546: Provide translated french translation on docs.python.org http://bugs.python.org/issue26546 #26539: frozen executables should have an empty path http://bugs.python.org/issue26539 #26536: Add the SIO_LOOPBACK_FAST_PATH option to socket.ioctl http://bugs.python.org/issue26536 #26535: Minor typo in the docs for struct.unpack http://bugs.python.org/issue26535 Top 10 most discussed issues (10) ================================= #26530: tracemalloc: add C API to manually track/untrack memory alloca http://bugs.python.org/issue26530 12 msgs #26512: Vocabulary: Using "integral" in library/stdtypes.html http://bugs.python.org/issue26512 8 msgs #26553: Write HTTP in uppercase http://bugs.python.org/issue26553 8 msgs #13305: datetime.strftime("%Y") not consistent for years < 1000 http://bugs.python.org/issue13305 7 msgs #26576: Tweak wording of decorator docos http://bugs.python.org/issue26576 7 msgs #22359: Remove incorrect uses of recursive make http://bugs.python.org/issue22359 4 msgs #22625: When cross-compiling, don’t try to execute binaries http://bugs.python.org/issue22625 4 msgs #23214: BufferedReader.read1(size) signature incompatible with Buffere http://bugs.python.org/issue23214 4 msgs #24959: unittest swallows part of stack trace when raising AssertionEr http://bugs.python.org/issue24959 4 msgs #26585: Use html.escape to replace _quote_html in http.server http://bugs.python.org/issue26585 4 msgs Issues closed (45) ================== #5505: sys.stdin.read() doesn't return after first EOF on Windows http://bugs.python.org/issue5505 closed by berker.peksag #16181: cookielib.http2time raises ValueError for invalid date. http://bugs.python.org/issue16181 closed by berker.peksag #17603: AC_LIBOBJ replacement of fileblocks http://bugs.python.org/issue17603 closed by martin.panter #17758: test_site fails when the user does not have a home directory http://bugs.python.org/issue17758 closed by haypo #18320: python installation is broken if prefix is overridden on an in http://bugs.python.org/issue18320 closed by ned.deily #20556: Use specific asserts in threading tests http://bugs.python.org/issue20556 closed by serhiy.storchaka #20589: pathlib.owner() and pathlib.group() raise ImportError on Windo http://bugs.python.org/issue20589 closed by berker.peksag #23606: ctypes.util.find_library("c") no longer makes sense http://bugs.python.org/issue23606 closed by steve.dower #23718: strptime() can produce invalid date with negative year day http://bugs.python.org/issue23718 closed by serhiy.storchaka #24918: Docs layout bug http://bugs.python.org/issue24918 closed by ezio.melotti #25320: unittest loader.py TypeError when code directory contains a so http://bugs.python.org/issue25320 closed by rbcollins #25638: Verify the etree_parse and etree_iterparse benchmarks are work http://bugs.python.org/issue25638 closed by brett.cannon #25687: Error during test case and tearDown http://bugs.python.org/issue25687 closed by ezio.melotti #25959: tkinter - PhotoImage.zoom() causes segfault http://bugs.python.org/issue25959 closed by terry.reedy #26079: Build with Visual Studio 2015 using PlatformToolset=v120 http://bugs.python.org/issue26079 closed by steve.dower #26102: access violation in PyErrFetch if tcur==null in PyGILState_Rel http://bugs.python.org/issue26102 closed by haypo #26247: Document Chrome/Chromium for python2.7 http://bugs.python.org/issue26247 closed by ezio.melotti #26313: ssl.py _load_windows_store_certs fails if windows cert store i http://bugs.python.org/issue26313 closed by steve.dower #26314: interned strings are stored in a dict, a set would use less me http://bugs.python.org/issue26314 closed by rhettinger #26323: Add assert_called() and assert_called_once() methods for mock http://bugs.python.org/issue26323 closed by berker.peksag #26499: http.client.IncompleteRead from HTTPResponse read after part r http://bugs.python.org/issue26499 closed by martin.panter #26500: Document of new star unpacking is missing. http://bugs.python.org/issue26500 closed by terry.reedy #26513: platform.win32_ver() broken in 2.7.11 http://bugs.python.org/issue26513 closed by steve.dower #26516: Add PYTHONMALLOC env var and add support for malloc debug hook http://bugs.python.org/issue26516 closed by haypo #26519: Cython doesn't work anymore on Python 3.6 http://bugs.python.org/issue26519 closed by haypo #26523: multiprocessing ThreadPool is untested http://bugs.python.org/issue26523 closed by pitrou #26538: regrtest: setup_tests() must not replace module.__path__ (_Nam http://bugs.python.org/issue26538 closed by haypo #26540: Python Macros http://bugs.python.org/issue26540 closed by ezio.melotti #26542: Wrongly formatted doctest block in difflib documentation http://bugs.python.org/issue26542 closed by berker.peksag #26548: Probably missing word in a sentence in the doc of bitwise oper http://bugs.python.org/issue26548 closed by rhettinger #26551: Regex.finditer infinite loops with certain input http://bugs.python.org/issue26551 closed by serhiy.storchaka #26555: string.format(bytes) raise warning http://bugs.python.org/issue26555 closed by haypo #26558: Disable PyGILState_Check() when Py_NewInterpreter() is called http://bugs.python.org/issue26558 closed by haypo #26561: exec function fails to properly assign scope to dict like obje http://bugs.python.org/issue26561 closed by belopolsky #26562: Large Involuntary context switches during oom-killer http://bugs.python.org/issue26562 closed by haypo #26563: PyMem_Malloc(): check that the GIL is hold in debug hooks http://bugs.python.org/issue26563 closed by haypo #26564: Malloc debug hooks: display memory block traceback on error http://bugs.python.org/issue26564 closed by haypo #26569: pyclbr.readmodule() and pyclbr.readmodule_ex() don't support n http://bugs.python.org/issue26569 closed by eric.snow #26570: comma-separated cookies with expires header do not parse prope http://bugs.python.org/issue26570 closed by SilentGhost #26572: urlparse does not handle passwords with ? in them. http://bugs.python.org/issue26572 closed by martin.panter #26573: Method Parameters can be Accepted as Keyword Arguments? http://bugs.python.org/issue26573 closed by G Young #26575: lambda not closed on specific value in comprehension http://bugs.python.org/issue26575 closed by martin.panter #26580: Documentation for ftplib still references ftpmirror.py http://bugs.python.org/issue26580 closed by berker.peksag #26583: test_timestamp_overflow of test_importlib fails if PYTHONDONTW http://bugs.python.org/issue26583 closed by ned.deily #747320: rfc2822 formatdate functionality duplication http://bugs.python.org/issue747320 closed by berker.peksag

1 0

Make the warnings module extensible
by Victor Stinner March 17, 2016

March 17, 2016

Hi, I have an API question for you. I would like to add a new parameter the the showwarning() function of the warnings module. Problem: it's not possible to do that without breaking the backward compatibility (when an application replaces warnings.showwarning(), the warnings allows and promotes that). I proposed a patch to add a new showmsg() function which takes a warnings.WarningMessage object: https://bugs.python.org/issue26568 The design is inspired by the logging module and its logging.LogRecord class. The warnings.WarningMessage already exists. Since it's class, it's easy to add new attributes without breaking the API. - If warnings.showwarning() is replaced by an application, this function will be called in practice to log the warning. - If warnings.showmsg() is replaced, again, this function will be called in practice. - If both functions are replaced, showmsg() will be called (replaced showwarning() is ignored) I'm not sure about function names: showmsg() and formatmsg(). Maybe: showwarnmsg() and formatwarnmsg()? Bikeshedding.... fight! The final goal is to log the traceback where the destroyed object was allocated when a ResourceWarning warning is logged: https://bugs.python.org/issue26567 Adding a new parameter to warnings make the implementation much more simple and gives more freedom to the logger to decide how to format the warning. Victor

1 0

Interested in the GSoC idea 'Roundup - GitHub integration'
by Wasim Thabraze March 16, 2016

March 16, 2016

Hello everyone, I am Wasim Thabraze, a Computer Science Undergraduate. I have thoroughly gone through the Core-Python GSoC ideas page and have narrowed down my choices to the project 'Improving Roundup GitHub integration'. I have experience in building stuff that are connected to GitHub. Openflock (http://www.openflock.co) is one of such products that I developed. Can someone please help me in knowing more about the project? I wanted to know how and where GitHub should be integrated in the https://bugs.python.org I hope I can code with Core Python this summer. Regards, Wasim www.thabraze.me github.com/waseem18

1 0

Really sketchy Core Python GSoC 2016 ideas page up
by Terri Oda March 16, 2016

March 16, 2016

Hey all, I've put up an incredibly sketchy, terrible ideas page up for Summer of Code with core python: https://wiki.python.org/moin/SummerOfCode/2016/python-core I'm pretty much the worst person to do this since I'm always swamped with admin stuff and particularly out of touch with what's needed in Core Python around this time of year, so I'm counting on you all to tell me that it's terrible and how to fix it. :) If you don't already have edit privileges on the python wiki and need them for this, let me know your wiki username and I can get you set up. (Or similarly, just tell me what needs fixing and make it my problem.) We're also still looking for more volunteers who can help mentor students. Let me know if this interests you -- we have a few folk who can do higher level code reviews, but we need more day-to-day mentors who can help students keep on track and help them figure out who to ask for help if they get stuck. I can also find folk who will provide mentoring for mentors if you'd like to try but don't think you could be a primary mentor without help! If you're interested in mentoring, email gsoc-admins(a)python.org and we can get you the link to sign up. And if you emailed before and didn't get a response or haven't found a group to work with, feel free to email again! Terri

1 0

Re: [Python-Dev] Bug in build system for cross-platform builds
by Martin Panter March 15, 2016

March 15, 2016

On 13 March 2016 at 01:13, Russell Keith-Magee <russell(a)keith-magee.com> wrote: > The patches that I've uploaded to Issue23670 [1] show a full cross-platform > [1] http://bugs.python.org/issue23670 > build process. After you apply that patch, the iOS directory contains a > meta-Makefile that manages the build process. Thanks very much for pointing that out. This has helped me understand a lot more things. Only now do I realize that the four files generated by pgen and _freeze_importlib are actually already committed into the Mercurial repository: Include/graminit.h Python/graminit.c Python/importlib.h Python/importlib_external.h A question for other Python developers: Why are these generated files stored in the repository? The graminit ones seem to have been there since forever (1990). It seems the importlib ones were there due to a bootstrapping problem, but now that is solved. Antoine <https://bugs.python.org/issue14928#msg163048> said he kept them in the repository on purpose, but I want to know why. If we ignore the cross compiling use case, would there be any other consequences of removing these generated files from the repository? E.g. would it affect the Windows build process? I have two possible solutions in mind: either remove the generated files from the repository and always build them, or keep them but do not automatically regenerate them every build. Since they are generated files, not source files, I would prefer to remove them, but I want to know the consequences first. > On Sat, Mar 12, 2016 at 8:48 AM, Martin Panter <vadmium+py(a)gmail.com> wrote: >> On 11 March 2016 at 23:16, Russell Keith-Magee <russell(a)keith-magee.com> >> wrote: >> > >> > On Sat, Mar 12, 2016 at 6:38 AM, Martin Panter <vadmium+py(a)gmail.com> >> > wrote: >> >> I don't understand. After I run Make, it looks like I get working >> >> executables leftover at Programs/_freeze_importlib and Parser/pgen. Do >> >> you mean to install these programs with "make install" or something? >> > >> > >> > Making them part of the installable artefacts would be one option, but >> > they >> > don't have to be installed, just preserved. >> >> What commands are you running that cause them to not be preserved at >> the end of the build? > > > I don't know - this is where I hit the end of my understanding of the build > process. All I know for certain is that 3.4.2 doesn't have this problem; > 3.5.1 does, and Issue22359 (fixed in [3]) is the source of the problem. A > subsequent fix [4] introduced an additional problem with _freeze_importlib. > > [3] https://hg.python.org/cpython/rev/565b96093ec8 > [4] https://hg.python.org/cpython/rev/02e3bf65b2f8 After my realization about the generated files, I think I can solve this one. Before the changes you identified, the build process probably thought the generated files were up to date, so it didn't need to cross-compile pgen or _freeze_importlib. If the generated file timestamps were out of date (e.g. depending on the order they are checked out or extracted), the first native build stage would have fixed them up.

6 11