|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] [PATCH RFC-for-4.12] tools/xen-foreign: Update python scripts to be Py3 compatible
On 3/5/19 12:30 PM, Juergen Gross wrote:
> On 05/03/2019 13:19, George Dunlap wrote:
>> On 3/5/19 12:04 PM, Wei Liu wrote:
>>> On Tue, Mar 05, 2019 at 12:41:53PM +0100, Juergen Gross wrote:
>>>> On 04/03/2019 20:00, George Dunlap wrote:
>>>>> On 3/4/19 6:31 PM, Andrew Cooper wrote:
>>>>>> The issues are:
>>>>>> * dict.has_key() was completely removed in Py3
>>>>>> * dict.keys() is an iterable rather than list in Py3, so .sort()
>>>>>> doesn't work.
>>>>>> * list.sort(cmp=) was deprecated in Py2.4 and removed in Py3. Replace
>>>>>> it
>>>>>> with a key= sort instead.
>>>>>>
>>>>>> This is all compatible with Py2.4 and later, which is when the sorted()
>>>>>> builtin was introduced. Tested with Py2.7 and Py3.4
>>>>>
>>>>> Did you test ./configure with python3?
>>>>>
>>>>> Loads of the python-related runes in the configure are also
>>>>> python2-only, resulting in (at least for me) the wrong library paths
>>>>> being generated, so the following fails:
>>>>>
>>>>> PYTHON=/usr/bin/python3 ./configure
>>>>>
>>>>> With the following message:
>>>>>
>>>>> File "<string>", line 1
>>>>> import distutils.sysconfig; print
>>>>> distutils.sysconfig.get_config_var("VERSION")
>>>>> ^
>>>>> SyntaxError: invalid syntax
>>>>> checking for python3-config... /home/gdunlap/anaconda3/bin/python3-config
>>>>> checking Python.h usability... yes
>>>>> checking Python.h presence... yes
>>>>> checking for Python.h... yes
>>>>> checking for PyArg_ParseTuple in -lpython... no
>>>>> configure: error: Unable to find a suitable python development library
>>>>> configure: error: ./configure failed for tools
>>>>>
>>>>> Note the error with the VERSION above; that results in looking for
>>>>> `-lpython` rather than `-lpython3.7`, which results in the following
>>>>> line being executed:
>>>>>
>>>>> configure:7468: checking for PyArg_ParseTuple in -lpython
>>>>> configure:7493: gcc -o conftest -g -O2 -g -O2
>>>>> -I/persistent/local/home/dunlapg/anaconda3/include/python3.7m
>>>>> -I/persistent/local/home/dunlapg/anaconda3/include/python3.7m
>>>>> -Wno-unused-result -Wsign-compare -march=nocona -mtune=haswell
>>>>> -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O3 -pipe
>>>>> -fdebug-prefix-map==/usr/local/src/conda/-
>>>>> -fdebug-prefix-map==/usr/local/src/conda-prefix -fuse-linker-plugin
>>>>> -ffat-lto-objects -flto-partition=none -flto -flto -fuse-linker-plugin
>>>>> -ffat-lto-objects -flto-partition=none -g -DNDEBUG -fwrapv -O3 -Wall
>>>>> -L/persistent/local/home/dunlapg/anaconda3/lib/python3.7/config-3.7m-x86_64-linux-gnu
>>>>> -L/persistent/local/home/dunlapg/anaconda3/lib -lpython3.7m -lpthread
>>>>> -ldl -lutil -lrt -lm -Xlinker -export-dynamic conftest.c -lpython
>>>>> -llzma -lpython3.7m -lpthread -ldl -lutil -lrt -lm -llzma >&5
>>>>> /usr/bin/ld: cannot find -lpython
>>>>> collect2: error: ld returned 1 exit status
>>>>>
>>>>> The following fixes it for me:
>>>>>
>>>>> diff --git a/tools/configure b/tools/configure
>>>>> index acc857510e..d7e21db959 100755
>>>>> --- a/tools/configure
>>>>> +++ b/tools/configure
>>>>> @@ -6953,7 +6953,7 @@ if echo "$PYTHON" | grep -q "^/"; then :
>>>>> PYTHON=`basename $PYTHONPATH`
>>>>>
>>>>> elif test -z "$PYTHON"; then :
>>>>> - PYTHON="python"
>>>>> + PYTHON="python2"
>>>>> else
>>>>> as_fn_error $? "PYTHON specified, but is not an absolute path"
>>>>> "$LINENO" 5
>>>>> fi
>>>>> diff --git a/tools/configure.ac b/tools/configure.ac
>>>>> index 1499344ce6..adf70aeb33 100644
>>>>> --- a/tools/configure.ac
>>>>> +++ b/tools/configure.ac
>>>>> @@ -355,7 +355,7 @@ esac
>>>>> AS_IF([echo "$PYTHON" | grep -q "^/"], [
>>>>> PYTHONPATH=$PYTHON
>>>>> PYTHON=`basename $PYTHONPATH`
>>>>> -],[test -z "$PYTHON"], [PYTHON="python"],
>>>>> +],[test -z "$PYTHON"], [PYTHON="python2"],
>>>>> [AC_MSG_ERROR([PYTHON specified, but is not an absolute path])])
>>>>> AX_PATH_PROG_OR_FAIL([PYTHONPATH], [$PYTHON])
>>>>> AX_CHECK_PYTHON_VERSION([2], [3])
>>>>>
>>>>> ..but:
>>>>> 1) I'm not sure if we can guarantee that `python2` will exist on all
>>>>> systems we support,
>>>>> 2) We probably want to support python3 at some point anyway.
>>>>>
>>>>> On the other hand, Arch has python -> python3, but (apparently) has
>>>>> since 2011 [1]; so there's an argument to be made that this isn't
>>>>> something that needs to be solved for 4.12.
>>>>
>>>> Which raises the question whether Andrew's patch should be included
>>>> in 4.12. I'm tempted to refuse my Rab as I don't see why we need to
>>>> be python3 compliant in the build system while configure isn't.
>>>>
>>>> Thoughts?
>>>
>>> I plan to have a closer look at python 3 compatibility issue today. But
>>> the outcome is unlikely to be simple -- most likely there will be a few
>>> patches. I don't think they will be suitable for 4.12.
>>>
>>> We can document the python 2 requirement for 4.12.
>>
>> Right, but there are two issues:
>> 1. Build system requires python2
>> 2. Build system assumes that 'python' == 'python2', and fails in strange
>> ways when it's not
>>
>> I agree #1 isn't something to rush before the release. Any opinions on #2?
>>
>> Options are:
>>
>> A. Simply document that if 'python' isn't 'python2' (such as on Arch
>> Linux, or if you've got anaconda python installed), you need to
>> explicitly set PYTHON=/path/to/python2
>>
>> B. Hard-code configure to look for python2 if not specified rather than
>> python.
>>
>> C. Have configure detect when python is python3, and look for python2,
>> failing if it's not available.
>>
>> Evaluation:
>>
>> Option A isn't a great UI, but the situation has been like this for a
>> long time, so it's probably not the worst thing ever to wait for the
>> next release to fix things. Also, serves you right for using such a
>> broken distro. :-)
>>
>> Option B is dead-simple; I just don't know how far back it will be
>> compatible.
>>
>> Option C is the most robust, but more work (though probably not a huge
>> amount).
>>
>> Thoughts?
>
> I'm hoping to have 4.12 finished rather soon (maybe one further RC?).
>
> For this reason I'd like to include fixes for regressions only from now
> on. This would imply option A.
Works for me.
-George
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel
|
![]() |
Lists.xenproject.org is hosted with RackSpace, monitoring our |