[Scons-dev] That was easy…

[Tim Jenness]

> On Dec 25, 2015, at 06:49 , Neal Becker <ndbecker2 at gmail.com> wrote:
> 
> Tim Jenness wrote:
> 
>> 
>>> On Dec 24, 2015, at 09:47 , Russel Winder <russel at winder.org.uk> wrote:
>>> 
>>> …well, sort of.
>>> 
>>> OK first major hurdle done: python3-port has now had default merged
>>> into it. Enough of it works to be able to compile Fortran code using
>>> this version of SCons on Python 2. Obviously (!) it all breaks using
>>> Python 3, and (obviously) many, many of the tests fail with any Python
>>> version.
>>> 
>>> There is now a long haul editing all the print statements into Python 3
>>> format in all the tests as well as the rest of the source code. I
>>> clearly need to find a way of doing this automatically. Aha 2-to-3 I
>>> think is required…
>>> 
>> 
>> I find that
>> 
>> futurize -1 *.py
>> 
>> is incredibly helpful here (-2 is the full on port to python 3 but -1 is
>> the command to make everything work with 2.7 features). This is from the
>> “future” package on PyPI.
>> 
>> I wouldn’t be surprised if the entire port to 2.7/3.3+ could have been
>> done on master using futurize directly (without six or a long-lived
>> porting branch).
>> 
>> —
>> Tim Jenness
>> 
> 
> Early on I really liked future package, but now it installs a ton of stuff 
> that causes massive conflicts, and I'm no longer a fan
> 

I’m not really sure what you mean. future package doesn’t have any dependencies as far as I can tell from the requirements.txt in the package. There are some additional dependencies if you need the code to work on python 2.6 but that’s not an issue here.

futurize -1 doesn’t inject dependencies into the code base itself.

— 
Tim Jenness