[python-committers] Do people prefer pushing feature repos or one massive patch?

Victor Stinner victor.stinner at gmail.com
Fri Apr 3 10:59:02 CEST 2015


Hi,

FYI the faulthandler and tracemalloc were both added in a single
commit, while they added a lot of new code and modified multiple
files.

The development of faulthandler and tracemalloc started as third party
projects on PyPI.

I almost rewrote tracemalloc from scratch while its PEP was discussed.
I didn't want to keep the history, because for such task (update the
implementation when the PEP changes), I use "hg commit" as I save a
file in an editor. I don't care of having well formed and atomic
changes. It's common to have a following "oops, fix ..." commit.


tracemalloc :

changeset:   87401:6e2089dbc5ad
user:        Victor Stinner <victor.stinner at gmail.com>
date:        Sat Nov 23 12:27:24 2013 +0100
files:       Doc/library/debug.rst Doc/library/tracemalloc.rst
Doc/license.rst Doc/using/cmd
description:
Issue #18874: Implement the PEP 454 (tracemalloc)

 Doc/library/debug.rst        |     1 +
 Doc/library/tracemalloc.rst  |   608 +++++++++++++++
 Doc/license.rst              |    41 +
 Doc/using/cmdline.rst        |    18 +-
 Lib/test/support/__init__.py |    19 +
 Lib/test/test_atexit.py      |     4 +-
 Lib/test/test_capi.py        |     2 +-
 Lib/test/test_threading.py   |     4 +-
 Lib/test/test_tracemalloc.py |   797 ++++++++++++++++++++
 Lib/tracemalloc.py           |   464 +++++++++++
 Modules/Setup.dist           |    17 +-
 Modules/_tracemalloc.c       |  1407 ++++++++++++++++++++++++++++++++++++
 Modules/hashtable.c          |   518 +++++++++++++
 Modules/hashtable.h          |   128 +++
 PC/config.c                  |     2 +
 PCbuild/pythoncore.vcxproj   |     5 +-
 Python/pythonrun.c           |     4 +
 17 files changed, 4024 insertions(+), 15 deletions(-)

faulthandler:

changeset:   69070:b0680b5a5215
user:        Victor Stinner <victor.stinner at haypocalc.com>
date:        Thu Mar 31 01:31:06 2011 +0200
files:       Doc/library/debug.rst Doc/library/faulthandler.rst Doc/usin
description:
Issue #11393: Add the new faulthandler module

 Doc/library/debug.rst         |    3 +-
 Doc/library/faulthandler.rst  |  129 ++++
 Doc/using/cmdline.rst         |    7 +
 Doc/whatsnew/3.3.rst          |    8 +
 Include/traceback.h           |   40 +
 Lib/test/regrtest.py          |    5 +
 Lib/test/script_helper.py     |    5 +-
 Lib/test/test_faulthandler.py |  469 ++++++++++++++++
 Misc/NEWS                     |    2 +
 Modules/Setup.dist            |    3 +
 Modules/faulthandler.c        |  971 ++++++++++++++++++++++++++++++++++
 Modules/main.c                |    1 +
 PC/config.c                   |    2 +
 PCbuild/pythoncore.vcproj     |    4 +
 Python/pythonrun.c            |   21 +
 Python/traceback.c            |  235 ++++++++
 configure                     |    2 +-
 configure.in                  |    2 +-
 pyconfig.h.in                 |    3 +
 19 files changed, 1907 insertions(+), 5 deletions(-)

Victor

2015-04-01 18:09 GMT+02:00 Brett Cannon <bcannon at gmail.com>:
> The implementation for PEP 488 is basically done (sans Windows installer
> stuff). I did the work in a features repo at
> https://hg.python.org/features/pep-488/ . Once I have addressed reviewer
> comments at http://bugs.python.org/issue23731 , would people prefer I simply
> push the features repo to hg.python.org/cpython and have the more granular
> history but have various "merge default" commits, or would people rather I
> do one massive commit?
>
> _______________________________________________
> python-committers mailing list
> python-committers at python.org
> https://mail.python.org/mailman/listinfo/python-committers
>


More information about the python-committers mailing list