| Submitter | Gregory Szorc |
|---|---|
| Date | Nov. 22, 2015, 6:14 a.m. |
| Message ID | <1101c57a493083f9fd3d.1448172846@ubuntu-main> |
| Download | mbox | patch |
| Permalink | /patch/11579/ |
| State | Accepted |
| Headers | show |
Comments
On 11/21/2015 10:14 PM, Gregory Szorc wrote: > # HG changeset patch > # User Gregory Szorc <gregory.szorc@gmail.com> > # Date 1448168629 28800 > # Sat Nov 21 21:03:49 2015 -0800 > # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a > # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b > hg: gracefully handle PyPy > > PyPy isn't compatible with C extensions nor the on demand module import > mechanism. First 4 pushed to the clowncopter. Nice rework. Check commit complains about find_module name, but I'll ignore that.
On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David < pierre-yves.david@ens-lyon.org> wrote: > > > On 11/21/2015 10:14 PM, Gregory Szorc wrote: > >> # HG changeset patch >> # User Gregory Szorc <gregory.szorc@gmail.com> >> # Date 1448168629 28800 >> # Sat Nov 21 21:03:49 2015 -0800 >> # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a >> # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b >> hg: gracefully handle PyPy >> >> PyPy isn't compatible with C extensions nor the on demand module import >> mechanism. >> > > First 4 pushed to the clowncopter. Nice rework. > > Check commit complains about find_module name, but I'll ignore that. There's nothing we can do about that: "find_module" is part of the import hook interface, which is defined by Python itself.
On 11/22/2015 08:09 PM, Gregory Szorc wrote: > On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David > <pierre-yves.david@ens-lyon.org <mailto:pierre-yves.david@ens-lyon.org>> > wrote: > > > > On 11/21/2015 10:14 PM, Gregory Szorc wrote: > > # HG changeset patch > # User Gregory Szorc <gregory.szorc@gmail.com > <mailto:gregory.szorc@gmail.com>> > # Date 1448168629 28800 > # Sat Nov 21 21:03:49 2015 -0800 > # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a > # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b > hg: gracefully handle PyPy > > PyPy isn't compatible with C extensions nor the on demand module > import > mechanism. > > > First 4 pushed to the clowncopter. Nice rework. > > Check commit complains about find_module name, but I'll ignore that. > > > There's nothing we can do about that: "find_module" is part of the > import hook interface, which is defined by Python itself. And that is why I'll ignore it ;-) However this series is breaking test-extensions.t when running from repository without local build (make local never ran). The pure finding logic apparently get confused when trying to generate __version__.py leading to the Mercurial to be isntall with -0+0 as a version number. Can you look into that and send me a followup quickly ?
On Sun, Nov 22, 2015 at 9:08 PM, Pierre-Yves David < pierre-yves.david@ens-lyon.org> wrote: > > > On 11/22/2015 08:09 PM, Gregory Szorc wrote: > >> On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David >> <pierre-yves.david@ens-lyon.org <mailto:pierre-yves.david@ens-lyon.org>> >> wrote: >> >> >> >> On 11/21/2015 10:14 PM, Gregory Szorc wrote: >> >> # HG changeset patch >> # User Gregory Szorc <gregory.szorc@gmail.com >> <mailto:gregory.szorc@gmail.com>> >> # Date 1448168629 28800 >> # Sat Nov 21 21:03:49 2015 -0800 >> # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a >> # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b >> hg: gracefully handle PyPy >> >> PyPy isn't compatible with C extensions nor the on demand module >> import >> mechanism. >> >> >> First 4 pushed to the clowncopter. Nice rework. >> >> Check commit complains about find_module name, but I'll ignore that. >> >> >> There's nothing we can do about that: "find_module" is part of the >> import hook interface, which is defined by Python itself. >> > > And that is why I'll ignore it ;-) > > However this series is breaking test-extensions.t when running from > repository without local build (make local never ran). The pure finding > logic apparently get confused when trying to generate __version__.py > leading to the Mercurial to be isntall with -0+0 as a version number. > > Can you look into that and send me a followup quickly ? I'm unable to reproduce. Do you have more exact STR? (I think I'm getting the run-tests.py arguments wrong.)
On 11/23/2015 11:16 AM, Gregory Szorc wrote: > On Sun, Nov 22, 2015 at 9:08 PM, Pierre-Yves David > <pierre-yves.david@ens-lyon.org <mailto:pierre-yves.david@ens-lyon.org>> > wrote: > > > > On 11/22/2015 08:09 PM, Gregory Szorc wrote: > > On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David > <pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org> > <mailto:pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org>>> > wrote: > > > > On 11/21/2015 10:14 PM, Gregory Szorc wrote: > > # HG changeset patch > # User Gregory Szorc <gregory.szorc@gmail.com > <mailto:gregory.szorc@gmail.com> > <mailto:gregory.szorc@gmail.com > <mailto:gregory.szorc@gmail.com>>> > # Date 1448168629 28800 > # Sat Nov 21 21:03:49 2015 -0800 > # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a > # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b > hg: gracefully handle PyPy > > PyPy isn't compatible with C extensions nor the on > demand module > import > mechanism. > > > First 4 pushed to the clowncopter. Nice rework. > > Check commit complains about find_module name, but I'll > ignore that. > > > There's nothing we can do about that: "find_module" is part of the > import hook interface, which is defined by Python itself. > > > And that is why I'll ignore it ;-) > > However this series is breaking test-extensions.t when running from > repository without local build (make local never ran). The pure > finding logic apparently get confused when trying to generate > __version__.py leading to the Mercurial to be isntall with -0+0 as a > version number. > > Can you look into that and send me a followup quickly ? > > > I'm unable to reproduce. Do you have more exact STR? (I think I'm > getting the run-tests.py arguments wrong.) cd tests hg purge --all ./run-tests.py test-extensions.y
On Mon, Nov 23, 2015 at 12:00 PM, Pierre-Yves David < pierre-yves.david@ens-lyon.org> wrote: > > > On 11/23/2015 11:16 AM, Gregory Szorc wrote: > >> On Sun, Nov 22, 2015 at 9:08 PM, Pierre-Yves David >> <pierre-yves.david@ens-lyon.org <mailto:pierre-yves.david@ens-lyon.org>> >> wrote: >> >> >> >> On 11/22/2015 08:09 PM, Gregory Szorc wrote: >> >> On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David >> <pierre-yves.david@ens-lyon.org >> <mailto:pierre-yves.david@ens-lyon.org> >> <mailto:pierre-yves.david@ens-lyon.org >> <mailto:pierre-yves.david@ens-lyon.org>>> >> wrote: >> >> >> >> On 11/21/2015 10:14 PM, Gregory Szorc wrote: >> >> # HG changeset patch >> # User Gregory Szorc <gregory.szorc@gmail.com >> <mailto:gregory.szorc@gmail.com> >> <mailto:gregory.szorc@gmail.com >> <mailto:gregory.szorc@gmail.com>>> >> # Date 1448168629 28800 >> # Sat Nov 21 21:03:49 2015 -0800 >> # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a >> # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b >> hg: gracefully handle PyPy >> >> PyPy isn't compatible with C extensions nor the on >> demand module >> import >> mechanism. >> >> >> First 4 pushed to the clowncopter. Nice rework. >> >> Check commit complains about find_module name, but I'll >> ignore that. >> >> >> There's nothing we can do about that: "find_module" is part of the >> import hook interface, which is defined by Python itself. >> >> >> And that is why I'll ignore it ;-) >> >> However this series is breaking test-extensions.t when running from >> repository without local build (make local never ran). The pure >> finding logic apparently get confused when trying to generate >> __version__.py leading to the Mercurial to be isntall with -0+0 as a >> version number. >> >> Can you look into that and send me a followup quickly ? >> >> >> I'm unable to reproduce. Do you have more exact STR? (I think I'm >> getting the run-tests.py arguments wrong.) >> > > cd tests > hg purge --all > ./run-tests.py test-extensions.y > I ran purge before, but not without --all. Looks like a .egg-info directory was interfering! I can reproduce this now. Looking at a fix.
On Mon, Nov 23, 2015 at 12:00 PM, Pierre-Yves David < pierre-yves.david@ens-lyon.org> wrote: > > > On 11/23/2015 11:16 AM, Gregory Szorc wrote: > >> On Sun, Nov 22, 2015 at 9:08 PM, Pierre-Yves David >> <pierre-yves.david@ens-lyon.org <mailto:pierre-yves.david@ens-lyon.org>> >> wrote: >> >> >> >> On 11/22/2015 08:09 PM, Gregory Szorc wrote: >> >> On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David >> <pierre-yves.david@ens-lyon.org >> <mailto:pierre-yves.david@ens-lyon.org> >> <mailto:pierre-yves.david@ens-lyon.org >> <mailto:pierre-yves.david@ens-lyon.org>>> >> wrote: >> >> >> >> On 11/21/2015 10:14 PM, Gregory Szorc wrote: >> >> # HG changeset patch >> # User Gregory Szorc <gregory.szorc@gmail.com >> <mailto:gregory.szorc@gmail.com> >> <mailto:gregory.szorc@gmail.com >> <mailto:gregory.szorc@gmail.com>>> >> # Date 1448168629 28800 >> # Sat Nov 21 21:03:49 2015 -0800 >> # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a >> # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b >> hg: gracefully handle PyPy >> >> PyPy isn't compatible with C extensions nor the on >> demand module >> import >> mechanism. >> >> >> First 4 pushed to the clowncopter. Nice rework. >> >> Check commit complains about find_module name, but I'll >> ignore that. >> >> >> There's nothing we can do about that: "find_module" is part of the >> import hook interface, which is defined by Python itself. >> >> >> And that is why I'll ignore it ;-) >> >> However this series is breaking test-extensions.t when running from >> repository without local build (make local never ran). The pure >> finding logic apparently get confused when trying to generate >> __version__.py leading to the Mercurial to be isntall with -0+0 as a >> version number. >> >> Can you look into that and send me a followup quickly ? >> >> >> I'm unable to reproduce. Do you have more exact STR? (I think I'm >> getting the run-tests.py arguments wrong.) >> > > cd tests > hg purge --all > ./run-tests.py test-extensions.y OK. This is a "fun" regression. Basically, setup.py is adding "mercurial/pure" to PYTHONPATH. This allows the `hg log` and `hg id` invocations performed as part of calculating the version string to work if no C extensions are present. This issue is fixed in the dropped part 5 by changing setup.py to perform the `hg` invocations with an import policy of "python only." Since this regression only seems to impact test-extension.t and rewriting the series to avoid the regression would involve a bit of work, is it acceptable if I send along a proper part 5 [as opposed to rewriting the series]? Feel free to drop this series from clowncopter in the interim.
On Mon, Nov 23, 2015 at 8:01 PM, Gregory Szorc <gregory.szorc@gmail.com> wrote: > On Mon, Nov 23, 2015 at 12:00 PM, Pierre-Yves David < > pierre-yves.david@ens-lyon.org> wrote: > >> >> >> On 11/23/2015 11:16 AM, Gregory Szorc wrote: >> >>> On Sun, Nov 22, 2015 at 9:08 PM, Pierre-Yves David >>> <pierre-yves.david@ens-lyon.org <mailto:pierre-yves.david@ens-lyon.org>> >>> wrote: >>> >>> >>> >>> On 11/22/2015 08:09 PM, Gregory Szorc wrote: >>> >>> On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David >>> <pierre-yves.david@ens-lyon.org >>> <mailto:pierre-yves.david@ens-lyon.org> >>> <mailto:pierre-yves.david@ens-lyon.org >>> <mailto:pierre-yves.david@ens-lyon.org>>> >>> wrote: >>> >>> >>> >>> On 11/21/2015 10:14 PM, Gregory Szorc wrote: >>> >>> # HG changeset patch >>> # User Gregory Szorc <gregory.szorc@gmail.com >>> <mailto:gregory.szorc@gmail.com> >>> <mailto:gregory.szorc@gmail.com >>> <mailto:gregory.szorc@gmail.com>>> >>> # Date 1448168629 28800 >>> # Sat Nov 21 21:03:49 2015 -0800 >>> # Node ID 1101c57a493083f9fd3dc2eb7832f862c587cd5a >>> # Parent b1f0ae9d3abf60a1326e305237c31ce3fee7262b >>> hg: gracefully handle PyPy >>> >>> PyPy isn't compatible with C extensions nor the on >>> demand module >>> import >>> mechanism. >>> >>> >>> First 4 pushed to the clowncopter. Nice rework. >>> >>> Check commit complains about find_module name, but I'll >>> ignore that. >>> >>> >>> There's nothing we can do about that: "find_module" is part of >>> the >>> import hook interface, which is defined by Python itself. >>> >>> >>> And that is why I'll ignore it ;-) >>> >>> However this series is breaking test-extensions.t when running from >>> repository without local build (make local never ran). The pure >>> finding logic apparently get confused when trying to generate >>> __version__.py leading to the Mercurial to be isntall with -0+0 as a >>> version number. >>> >>> Can you look into that and send me a followup quickly ? >>> >>> >>> I'm unable to reproduce. Do you have more exact STR? (I think I'm >>> getting the run-tests.py arguments wrong.) >>> >> >> cd tests >> hg purge --all >> ./run-tests.py test-extensions.y > > > OK. This is a "fun" regression. > > Basically, setup.py is adding "mercurial/pure" to PYTHONPATH. This allows > the `hg log` and `hg id` invocations performed as part of calculating the > version string to work if no C extensions are present. > > This issue is fixed in the dropped part 5 by changing setup.py to perform > the `hg` invocations with an import policy of "python only." > > Since this regression only seems to impact test-extension.t and rewriting > the series to avoid the regression would involve a bit of work, is it > acceptable if I send along a proper part 5 [as opposed to rewriting the > series]? Feel free to drop this series from clowncopter in the interim. > > Actually, please drop the entire series: I need to rewrite a clowncoptered patch to avoid a worse regression.
On 11/23/2015 08:22 PM, Gregory Szorc wrote: > On Mon, Nov 23, 2015 at 8:01 PM, Gregory Szorc <gregory.szorc@gmail.com > <mailto:gregory.szorc@gmail.com>> wrote: > > On Mon, Nov 23, 2015 at 12:00 PM, Pierre-Yves David > <pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org>> wrote: > > > > On 11/23/2015 11:16 AM, Gregory Szorc wrote: > > On Sun, Nov 22, 2015 at 9:08 PM, Pierre-Yves David > <pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org> > <mailto:pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org>>> > wrote: > > > > On 11/22/2015 08:09 PM, Gregory Szorc wrote: > > On Sun, Nov 22, 2015 at 7:45 PM, Pierre-Yves David > <pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org> > <mailto:pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org>> > <mailto:pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org> > <mailto:pierre-yves.david@ens-lyon.org > <mailto:pierre-yves.david@ens-lyon.org>>>> > wrote: > > > > On 11/21/2015 10:14 PM, Gregory Szorc wrote: > > # HG changeset patch > # User Gregory Szorc > <gregory.szorc@gmail.com <mailto:gregory.szorc@gmail.com> > <mailto:gregory.szorc@gmail.com > <mailto:gregory.szorc@gmail.com>> > <mailto:gregory.szorc@gmail.com > <mailto:gregory.szorc@gmail.com> > <mailto:gregory.szorc@gmail.com > <mailto:gregory.szorc@gmail.com>>>> > # Date 1448168629 28800 > # Sat Nov 21 21:03:49 2015 -0800 > # Node ID > 1101c57a493083f9fd3dc2eb7832f862c587cd5a > # Parent > b1f0ae9d3abf60a1326e305237c31ce3fee7262b > hg: gracefully handle PyPy > > PyPy isn't compatible with C extensions > nor the on > demand module > import > mechanism. > > > First 4 pushed to the clowncopter. Nice rework. > > Check commit complains about find_module name, > but I'll > ignore that. > > > There's nothing we can do about that: "find_module" > is part of the > import hook interface, which is defined by Python > itself. > > > And that is why I'll ignore it ;-) > > However this series is breaking test-extensions.t when > running from > repository without local build (make local never ran). > The pure > finding logic apparently get confused when trying to > generate > __version__.py leading to the Mercurial to be isntall > with -0+0 as a > version number. > > Can you look into that and send me a followup quickly ? > > > I'm unable to reproduce. Do you have more exact STR? (I > think I'm > getting the run-tests.py arguments wrong.) > > > cd tests > hg purge --all > ./run-tests.py test-extensions.y > > > OK. This is a "fun" regression. > > Basically, setup.py is adding "mercurial/pure" to PYTHONPATH. This > allows the `hg log` and `hg id` invocations performed as part of > calculating the version string to work if no C extensions are present. > > This issue is fixed in the dropped part 5 by changing setup.py to > perform the `hg` invocations with an import policy of "python only." > > Since this regression only seems to impact test-extension.t and > rewriting the series to avoid the regression would involve a bit of > work, is it acceptable if I send along a proper part 5 [as opposed > to rewriting the series]? Feel free to drop this series from > clowncopter in the interim. > > > Actually, please drop the entire series: I need to rewrite a > clowncoptered patch to avoid a worse regression. Series have been properly obliterated in the clowncopter.
On Sun, Nov 22, 2015 at 1:14 AM, Gregory Szorc <gregory.szorc@gmail.com> wrote: > +if '__pypy__' in sys.builtin_module_names: > + pypy = True > +else: > + pypy = False > > pypy = '__pypy__' in sys.builtin_module_names
Patch
diff --git a/hg b/hg --- a/hg +++ b/hg @@ -10,48 +10,62 @@ import imp import os import sys if os.environ.get('HGUNICODEPEDANTRY', False): reload(sys) sys.setdefaultencoding("undefined") +# PyPy doesn't load C extensions and isn't compatible with demand importing. +# +# The canonical way to do this is to test platform.python_implementation(). +# But we don't import platform anywhere, so don't require it here. +if '__pypy__' in sys.builtin_module_names: + pypy = True +else: + pypy = False libdir = '@LIBDIR@' # Rules for how modules can be loaded. Values are: # c - require C extensions # allow - allow pure Python implementation when C loading fails # py - only load pure Python modules modulepolicy = '@MODULELOADPOLICY@' # By default, require the C extensions for performance reasons. if modulepolicy == '@' 'MODULELOADPOLICY' '@': modulepolicy = 'c' +# PyPy isn't compatible with C extensions. So don't try to load them. +if pypy: + modulepolicy = 'py' + # Environment variable can always force settings. modulepolicy = os.environ.get('HGMODULEPOLICY', modulepolicy) if libdir != '@' 'LIBDIR' '@': if not os.path.isabs(libdir): libdir = os.path.join(os.path.dirname(os.path.realpath(__file__)), libdir) libdir = os.path.abspath(libdir) sys.path.insert(0, libdir) -# enable importing on demand to reduce startup time -try: - from mercurial import demandimport; demandimport.enable() -except ImportError: - import sys - sys.stderr.write("abort: couldn't find mercurial libraries in [%s]\n" % - ' '.join(sys.path)) - sys.stderr.write("(check your install and PYTHONPATH)\n") - sys.exit(-1) +# On demand importing doesn't work with PyPy (yet). +if not pypy: + # Enable importing on demand to reduce startup time. + try: + from mercurial import demandimport; demandimport.enable() + except ImportError: + import sys + sys.stderr.write("abort: couldn't find mercurial libraries in [%s]\n" % + ' '.join(sys.path)) + sys.stderr.write("(check your install and PYTHONPATH)\n") + sys.exit(-1) # Install a PEP-302 custom module finder and loader that knows how to # import modules with implementations in both Python and C. # List of modules that have both Python and C implementations. See also the # set of .py files under mercurial/pure/. dualmodules = set([ 'mercurial.base85',