[PATCH 1/5] debian/control: build-depend on python3-sphinx instead of python-sphinx

[Daniel Kahn Gillmor]

On Sun 2017-12-03 15:35:04 +0200, Tomi Ollila wrote:
> On Thu, Nov 30 2017, Daniel Kahn Gillmor wrote:
>
>> On Thu 2017-11-30 04:40:39 -0500, Daniel Kahn Gillmor wrote:
>>> python2 is going to be deprecated, and python3-sphinx is available all
>>> the way back to oldoldstable.  let's use the more modern version.
>>> ---
>>>  debian/control | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/debian/control b/debian/control
>>> index 20b8a2db..3a624fdc 100644
>>> --- a/debian/control
>>> +++ b/debian/control
>>> @@ -18,7 +18,7 @@ Build-Depends:
>>>   python3-all (>= 3.1.2-7~),
>>>   dh-python,
>>>   dh-elpa (>= 1.3),
>>> - python-sphinx (>= 1.0),
>>> + python3-sphinx,
>>>   ruby, ruby-dev (>>1:1.9.3~),
>>>   emacs25-nox | emacs25 (>=25~) | emacs25-lucid (>=25~) |
>>>   emacs24-nox | emacs24 (>=24~) | emacs24-lucid (>=24~) |
>>
>> hm, on further reflection, this one is specifically problematic because
>> of the use of:
>>
>>     ${python} -m sphinx.writers.manpage
>>
>> in ./configure, used to test for the presence of the sphinx module.
>>
>> This can be fixed by preferring python3 here:
>>
>> --- a/configure
>> +++ b/configure
>> @@ -557,7 +557,7 @@ fi
>>  printf "Checking for python... "
>>  have_python=0
>>  
>> -for name in ${PYTHON} python python2 python3; do
>> +for name in ${PYTHON} python3 python python2; do
>
>>      if command -v $name > /dev/null; then
>>         have_python=1
>>         python=$name
>>
>> any thoughts or preferences on this?
>
> I'd like this change. I was going to suggest "${PYTHON}" there
> but perhaps if one sets e.g.
>
> PYTHON='/usr/local/bin/python3 -B -E -u'
>
> The above would still work (I'd say this is potentially more useful
> than supporting whitespace in paths).

does it work?  i think the -B -E -u will all be tried as possible python
implementations separately, no?  and $python itself will just get set to
/usr/local/bin/python3 with the arguments dropped :(

0 dkg at alice:~$ x="a b c"
0 dkg at alice:~$ for y in ${x} d e; do printf ':%s\n' "$y"; done
:a
:b
:c
:d
:e
0 dkg at alice:~$ 

While i think this doesn't support Tomi's extended use case, i still
think it's the right direction to move in.

Does anyone have any objection to the preference order change above?  Is
there any reason we should be preferring python 2 to python3 as we move
forward?

        --dkg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20171204/91641ec6/attachment.sig>