[Mobile-sig] [Python-Dev] Import Fails in setup.py On Android

Cyd Haselton chaselton at gmail.com
Mon Feb 16 00:28:45 CET 2015 

After banging my head against libfakechroot for days I've finally managed to fix it to work on Android. 

Moving back to Python...I ran make clean && make && make install. I run the installed python3 binary, run help() and get the following error:

ype "help", "copyright", "credits" or "license" for more information.
>>> help()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/python/lib/python3.4/_sitebuiltins.py", line 102, in __call__
    import pydoc
  File "/usr/python/lib/python3.4/pydoc.py", line 62, in <module>
    import platform
  File "/usr/python/lib/python3.4/platform.py", line 115, in <module>
    import sys, os, re, subprocess
  File "/usr/python/lib/python3.4/subprocess.py", line 406, in <module>
    import select
ImportError: No module named 'select'


I then attempt to import select, then ctypes, and get the following errors:
>>> import select
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: No module named 'select'
>>> import ctypes
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/python/lib/python3.4/ctypes/__init__.py", line 22, in <module>
    from struct import calcsize as _calcsize
  File "/usr/python/lib/python3.4/struct.py", line 13, in <module>
    from _struct import *
ImportError: dlopen failed: cannot locate symbol "PyLong_AsSize_t" referenced by "_struct.cpython-34m.so"...
>>> import _struct
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: dlopen failed: cannot locate symbol "PyLong_AsSize_t" referenced by "_struct.cpython-34m.so"...
>>>

Googling the ImportError turns up six results...none of them related to a missing PyLong_AsSize_t symbol.

Has anyone seen this error before?

Thanks in advance.  Headed back to look at the _struct code.

Cyd

On February 8, 2015 6:46:02 PM CST, Cyd Haselton <chaselton at gmail.com> wrote:
>Quick Update:
>
>I've managed to re-compile a libfakechroot that (I believe) eliminates
>the 'undefined reference to dlopen()' issues; problems importing
>extensions during the build process now display meaningful errors
>instead of crashing the entire thing.
>
>Some imports failed because of missing symbols (i.e. _crypt, which
>makes sense because I don't have -lcrypt on device) and some failed
>because of linking issues (i.e _ctypes).  The majority failed to
>import due to an undeclared variable in setup.py...which was quickly
>fixed.
>
>Once the most recent build attempt is done, I'll post relevant stack
>traces and errors, if any.
>
>Cyd
>
>On Sat, Feb 7, 2015 at 5:36 AM, Cyd Haselton <chaselton at gmail.com>
>wrote:
>> Hello Russ,
>> Thanks for the update.
>>
>> I'm building Python 3.4.2 from sources, (using a GCC that I
>originally
>> ported, then upgraded on device) on an Android device in an
>environment that
>> simulates a Linux filesystem using fakechroot (Terminal Emulator and
>KBOX);
>> Python is not currently installed in this environment.
>>
>> I've built other utilities in this environment successfully; the key
>is
>> making sure that anything that requires dlopen() is built with shared
>> libraries and is linked with -lc -ldl, as Android's dlopen() is not
>in its
>> libc but in -ldl. The shared part is due to a limitation of the
>> libfakechroot that the KBOX environment uses; anything calling
>dlopen()
>> without the necessary links will cause fakechroot to terminate the
>process
>> and throw an undefined symbol error
>>
>> For the Python build I've specified -lc -ldl in the appropriate
>> environmental variables, in Modules/Setup and in setup.py. The
>initial
>> python binary is built, but it fails (with the undefined ref to
>dlopen())
>> when importing the extension modules it has has built. I have run nm
>on
>> those modules and...even though -lc -ldl is on the extension build
>command
>> line...the dlopen() symbol is not there.
>>
>> Sorry for the long explanation. Let me know if there's info I've left
>out.
>>
>>
>>
>> On February 6, 2015 8:10:16 PM CST, Russell Keith-Magee
>> <russell at keith-magee.com> wrote:
>>>
>>> Hi Cyd,
>>>
>>> This list is as good a place as any for what you're describing.
>>>
>>> However, it's a little difficult to provide advice without stack
>traces
>>> and details of how you've built your Python build to start with. For
>>> example, are you cross compiling? If so, the problem you've
>described could
>>> be caused by referencing the wrong Python instance when running
>setup.py.
>>>
>>> Yours,
>>> Russ Magee %-)
>>>
>>> On Fri, Feb 6, 2015 at 3:46 AM, Cyd Haselton <chaselton at gmail.com>
>wrote:
>>>>
>>>> Also, if I've got the wrong list, please let me know.  Links to the
>>>> correct list/lists are optional; I just want to know if this is the
>>>> right place to get help so that I can keep searching if it isn't.
>>>>
>>>> On Thu, Feb 5, 2015 at 5:40 AM, Cyd Haselton <chaselton at gmail.com>
>wrote:
>>>> > Since i'm working on an Android tablet, I am forwarding the
>original
>>>> > message
>>>> > I sent to python-dev instead of re-typing the entire thing.
>>>> >
>>>> > To summarize: I'm building in a fakechroot environment on Android
>that
>>>> > supports dynamic calls to libc only. Something in the
>>>> > importlib._bootstrap
>>>> > process is not linking to...or has not been linked to...the
>necessary
>>>> > libraries (I've checked the module...it is being linked to -lc
>-ldl)
>>>> > but I
>>>> > can't figure out what...as there's not a lot in the python docs
>on it.
>>>> >
>>>> > I'm currently reading through the importlib docs but any help
>would be
>>>> > greatly appreciated.
>>>> >
>>>> > ________________________________
>>>> > From: Eric Snow <ericsnowcurrently at gmail.com>
>>>> > Sent: February 3, 2015 12:10:54 AM CST
>>>> > To: Cyd Haselton <chaselton at gmail.com>
>>>> > Cc: Python-Dev <python-dev at python.org>
>>>> > Subject: Re: [Python-Dev] Import Fails in setup.py On Android
>>>> >
>>>> > On Mon, Feb 2, 2015 at 12:36 PM, Cyd Haselton
><chaselton at gmail.com>
>>>> > wrote:
>>>> >>
>>>> >>  After fixing a segfault issue (many thanks Ryan) I'm back to
>the same
>>>> >> issue
>>>> >>  I was having with Python 2.7.8; the newly built python throws
>an
>>>> >> undefined
>>>> >>  reference to dlopen when running setup.py...specifically when
>>>> >> importing
>>>> >>  just-built extensions
>>>> >>
>>>> >>  I've managed to narrow the problem down to the following line:
>>>> >>
>>>> >>  importlib._bootstrap._SpecMethods(spec).load()
>>>> >
>>>> >
>>>> > That call is where modules are created (and executed) via the
>loader
>>>> > designated for the module by the import machinery.  So a problem
>here
>>>> > may simply reflect a problem with the loader.  Which module is
>being
>>>> > imported when you run into trouble?  Is it a C-extension module?
>>>> >
>>>> > Also keep in mind that
>>>> > basically everything in importlib._bootstrap is
>>>> > frozen into your Python binary when you build Python (or should
>be).
>>>> > So if you are seeing any errors related to something missing or
>broken
>>>> > in importlib._bootstrap, it could be related to the build step
>where
>>>> > importlib gets frozen.
>>>> >
>>>> > Either way, more info (e.g. a traceback) would be great if you
>need
>>>> > more
>>>> > help.
>>>> >
>>>> > -eric
>>>> >
>>>> >
>>>> > --
>>>> > Sent from my Android device with K-9 Mail. Please excuse my
>brevity.
>>>> _______________________________________________
>>>> Mobile-sig mailing list
>>>> Mobile-sig at python.org
>>>> https://mail.python.org/mailman/listinfo/mobile-sig
>>>
>>>
>>
>> --
>> Sent from my Android device with K-9 Mail. Please excuse my brevity.

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/mobile-sig/attachments/20150215/5a99cd63/attachment.html>

More information about the Mobile-sig mailing list