On Oct 24, 2013, at 2:15 PM, Satish Balay <[email protected]> wrote:

> An emphirical count on my recent mail archives.. I'm not sure if grep
> is able to access all configure.log files in these mailboxes - but the
> ratio could be representative..
> 
> so 10% of logs found are using python 2.4/2.5

  What about in the past year? 

   And how come you only have 70 configure.log we must have received far more 
than that?

   Barry

> 
> Satish
> 
> -----------
> 
> balay@mockingbird /home/balay/mail
> $ grep --no-group-separator -A 1 'Python version' inbox.old.39 inbox.old.41 
> inbox.old.40 |grep -v 'Python'  | wc -l
> 70
> balay@mockingbird /home/balay/mail
> $ grep --no-group-separator -A 1 'Python version' inbox.old.39 inbox.old.41 
> inbox.old.40 |grep -v 'Python'  | egrep "(2\.4|2\.5)" | wc -l
> 7
> 
> On Thu, 24 Oct 2013, Barry Smith wrote:
> 
>> 
>> configure.log contains the python version used. Can we scarf up the version 
>> for say the last three years of all configure.log we have received and view 
>> the trend of pre 2.6 ones still being used? Won’t be a perfect measure, but 
>> at least it is a measure.
>> 
>>   Barry
>> 
>> On Oct 24, 2013, at 1:52 PM, Jed Brown <[email protected]> wrote:
>> 
>>> Satish Balay <[email protected]> writes:
>>> 
>>>> Do you know if archlinux will switch over to using the guideline from 
>>>> python.org?
>>> 
>>> The guideline does not prohibit Arch from doing what they did (3 years
>>> ago).  It says that scripts should use "python2" if they will only work
>>> for python-2.x and use "python" if they work for both python-2.x (x=6 or
>>> 7 in practice) and 3.y.
>>> 
>>>> We are currently using python version from RHEL as a guideline. RHEL5
>>>> has python 2.4 with eol in march-2017.
>>>> 
>>>> And I see RHEL6 has python-2.6
>>> 
>>> Python-2.5 was released in 2006, so this is more than 10 years.  We're
>>> not very tolerant of PETSc users that are still using petsc-2.1.2.
>>> 
>>> This is not to say we have to drop support right away, but python-2.4 is
>>> getting quite old and forces us to use a number of more fragile
>>> constructs.  We may have trouble holding out until 2017.
>> 
>> 

Reply via email to