Comments on: It’s time for Python 2.6

By: eliben

eliben — Fri, 14 Aug 2009 04:52:59 +0000

@José,

Think of a named tuple like a once-created-then-accessed Struct. It’s convenient to pass complex arguments or return values around. I guess dicts can also be used for that, but less naturally. And of course dicts are useful for other things as well.

By: José Hérnandez

José Hérnandez — Wed, 12 Aug 2009 12:35:54 +0000

Thanks for the excellent review of 2.6! One question did arise after reading about namedtuples: what’s the difference between that and dictionaries?

By: eliben

eliben — Wed, 12 Aug 2009 03:14:01 +0000

@Michael,

My point, I guess, is just to say that except a few “language lawyer nitpicks”, YAML is a superset of JSON. So just including YAML would give JSON-lovers their JSON and YAML-lovers their extended features.

By: Michael Foord

Michael Foord — Sun, 09 Aug 2009 19:48:56 +0000

JSON is much more widely used than YAML.

By: Marcin Cieslik

Marcin Cieslik — Sat, 08 Aug 2009 13:04:56 +0000

Why JSON/YAML? Well I don’t and cPickle, marshal are certainly better (but not always faster!). But users of my software might want a text format. The software (PaPy) uses serialization to exchange data between processes in a parallel pipeline. I wanted to make it protocol agnostic, but YAML failed miserably (it scales worse then linear!), while JSON does not. To avoid user problems I dropped YAML.

By: eliben

eliben — Sat, 08 Aug 2009 05:03:42 +0000

@Marcin,

Out of curiosity – why would you want to serialize a 100MB file as JSON/YAML? These are text formats, and as such aren’t suitable for such large scale serialization. Wouldn’t cPickle or shelve be better for this?

By: René A.

René A. — Fri, 07 Aug 2009 16:36:26 +0000

Talking about convenience… also, besides multiple Python installations, really give ‘virtualenv’ a try.

You can customize on a per project basis you’re desired python environment, ranging from your Python version to a minimalistic set of desired packages.

By: Marcin Cieslik

Marcin Cieslik — Fri, 07 Aug 2009 12:45:32 +0000

The performance of YAML is really too painful to be a substitute for JSON. Try serializing/de-serializing some 100Mb (I did not succeed). Embarrassing, I have priority Queue in one of my codes:
class PriorityQueue(Queue): """ A priority queue using a heap on a list. This Queue is thread but not process safe. """ def _init(self, maxsize): self.maxsize = maxsize self.queue = []


    def _put(self, item):

        return heappush(self.queue, item)

def _get(self): return heappop(self.queue)
I guess I should read the docs more carefully

By: Eric Olivier LEBIGOT (EOL)

Eric Olivier LEBIGOT (EOL) — Fri, 07 Aug 2009 10:58:15 +0000

Nice overview of why Python 2.6 is a great upgrade!

Why did you have to uninstall Python 2.5 modules? Under Mac OS X, I have co-existing Python 2.5 and Python 2.6 installs, complete with 3rd-party modules. This is really convenient when you _have_ to use Python 2.5.