Hi all-
First, thanks for the great tool. I've been using Gentoo for about 10
years, and I use Catalyst to build custom Gentoo live CDs that contain
extra software (e.g. netcat6, rsync, ddrescue, etc.).
At some point in the past the stage3 tarballs only contained Python 3,
though it looks like the 20130516 amd64 stage3 contains both 2.7 and
3.2. It's reasonable to assume that any current Gentoo system has a
Python 3.2+ interpreter, so is there any objection to making Catalyst
work on Python 3? The fewer things in the world that depend on Python 2,
the better (in my opinion). Portage has supported Python 3 for some time
and it makes sense for Catalyst to follow suit.
I use Python 3 for all of my own work and have a reasonable amount of
experience porting projects to Python 3. This typically involves a lot
of code churn even when there aren't any logic changes -- every print
statement has to be converted to calling the print function, for
instance. Explicit conversion between bytes and str can require a bit
more thought but isn't difficult.
This conversion might require other miscellaneous cleanup, including
some whitespace-only commits. Python 3 does not allow mixing tabs and
spaces for indentation, and (e.g.) catalyst_support.py is a *mess* with
indentation. PyCharm also notices things like unnecessary "\" line
continuation characters, unused variables, mutable default arguments
(e.g. catalyst_support.spawn's "env"), equality comparisons against
None, and unreachable code (lines 183 - 220 of catalyst_lock.py).
It looks like the project has mostly standardized on using tabs for
indentation, so that's what I'll use when changing indentation is necessary.
In any case, over the next few weeks/months I'll prepare a series of
patches in decreasing order of priority:
1. Changes required to run 2to3, like changing string exceptions (e.g.
catalyst_lock.py:34) to 'raise Exception("message")'
2. Automated 2to3 conversion
3. Minimal indentation changes required for Python 3 to parse the code
(as I mentioned, the diff of catalyst_support.py will still be quite
large)
4. Any required bytes/str fixes and other things that 2to3 isn't smart
enough to do on its own
5. Miscellaneous style tweaks (some of which are suggested by PEP 8), like:
1. using os.path.join where appropriate (e.g. grp_target.py:38,
catalyst_lock.py:80)
2. using context managers ("with open(filename) as f:")
3. replacing mutable default function arguments
4. using argparse instead of getopt in the main catalyst script (maybe)
5. catching Exception instead of using a raw 'except:' -- probably
don't want to catch SystemExit, StopIteration and
KeyboardInterrupt in a lot of these places
6. Other PEP 8-inspired changes like using spaces around operators and
function arguments
The intention is that categories 1 through 4 would be the minimum set of
changes required to make Catalyst run on Python 3. Categories 5 and 6
would be a mix of PEP 8 and newer Python idioms and applying these would
be optional.
I'm not a Gentoo developer and I'll be making a lot of small commits, so
sharing my changes will probably be easiest by publishing a clone of the
Catalyst repository on GitHub. (I plan to make/test/release all of these
changes whether or not they'll make it in to Catalyst in the near
future. Even if the Gentoo releng team doesn't care about supporting
Python 3 now, having a working port available might make it easier to
justify that change in the future :) )
Thanks for your time and any feedback,
MMR...
--
Matthew Ruffalo
Department of Electrical Engineering and Computer Science
Case Western Reserve University
