Comments on: Python 3 and Growth (or the lack thereof)

By: Python, Growth, and Sandboxes | cemerick

Python, Growth, and Sandboxes | cemerick — Sun, 01 Apr 2007 04:04:23 +0000

[...] ← Python 3 and Growth (or the lack thereof) Thoughts on Martin Fowler’s Domain Specific Languages Overview [...]

By: Chas Emerick

Chas Emerick — Thu, 29 Mar 2007 03:56:57 +0000

I think Python could be and do amazing things. However, that’s not up to me, unless I were to become a full-time Python hacker (and even then, the most well-respected Python hackers aren’t GvR). “Huge growth” almost always requires breaking stuff from the past — and in this case, breaking founding principles as well. This is especially true with Python’s adherence to the principle of There’s Only One Way To Do It — the finest examples of metaprogramming facilities always exhibit the opposite, which is a sign of expressiveness and flexibility. (Of course, if one makes a few “bad” decisions, expressiveness and flexibility can lead to line noise. It’s a fine balance; I do not envy language designers.)

By: Chas Emerick

Chas Emerick — Thu, 29 Mar 2007 03:56:28 +0000

I forced myself to unsubscribe from python-dev some months back when I realized that Python and I were on different paths. So many times, I had knocked out a paragraph or so of rebuttal to some decision, but then realized (thankfully in time!) that I wasn’t a contributor, didn’t know the source, and would simply be pissing into the wind that I still continue to reap. Mixed metaphors there, but you get my drift.

I don’t really think it’s that grim in an absolute sense. Python is still going to kick butt for years to come in a variety of fields; web programming, hard science, misc. research, etc. will keep it going strong for a long, long time. And, for those of us who look to deliver value outside of Python strongholds, there’s tons of innovations and innovative languages and platforms elsewhere.

By: Doug

Doug — Thu, 29 Mar 2007 03:54:46 +0000

Guido is wrong, it is possible. It’s been done already: http://boo.codehaus.org/Closures

By: Meek

Meek — Thu, 29 Mar 2007 03:51:32 +0000

Python is not growing because you want programmable syntax and “esoteric” features? Features that 99% of software developers should never use. Let me guess, you have never maintained a project written in a language that supports programmable syntax where geniuses abuse meta-programming where simpler alternatives achieve the same goal.

Python has areas it can grow in (e.g performance, better memory management, better profiling/diagnostic tools, unified/consistent library, smarter virtual machine (see PyPy), cruft removal, better/easier extension development in C/C++, etc), but adding new abstractions or syntatic sugar that one in one million programmers know how and when to use is not an area that deserves Python’s attention. We have Ruby and Perl for that.

By: Phillip J. Eby

Phillip J. Eby — Thu, 29 Mar 2007 03:50:48 +0000

Um, so you don’t think the “with” statement and coroutines were new features?

What about the new metaclass hook that’ll be in Python 3.0 (and maybe 2.6)? It’s actually a pretty significant step forward for implementing Ruby-like DSL’s in Python.

Meanwhile, if somebody could come up with a Python-like way of implementing blocks, Guido might actually go with it. It’s just that at this point, he doesn’t want bike-shed discussions about it, or assertions that it must be possible to have such a syntax, as he’s convinced that it’s not possible/practical. Someone would have to come up with a concrete proposal that perhaps handles the problem in an entirely new way.

By: Tim D

Tim D — Thu, 29 Mar 2007 03:50:21 +0000

As fun as new language features are, I think the useful innovations come in the form of new libraries. Nice syntax makes my job more pleasant, but it is the libraries which actually help productivity.

By: Mike Bayer

Mike Bayer — Thu, 29 Mar 2007 03:49:31 +0000

to me, i think whats “hot” in a language is when it stops adding new syntaxes and tricks, continues to remove its older and less-polished paradigms and converges on a limited set of absolutely perfect constructs from which I can express anything. Python was exactly the language that had this characteristic which is why I chose it.

if you need a constant flow of newly released syntactic sugars and meta-trickery in order to feel like a language will allow you to “solve a problem”, that sounds like a need for fewer silver bullets and more target practice.

By: Paul Boddie

Paul Boddie — Thu, 29 Mar 2007 03:49:00 +0000

Why do you think languages need to continuously grow new features? Take the current DSL hype as an example: a lot of it is just people who don’t know how to write their own parsers and want to use their favourite programming language’s parser to prop up their back-of-the-napkin notion of what they should be writing instead of the boilerplate they’d otherwise need to write, all whilst ignoring existing languages designed for the very same purposes. Although I don’t doubt that good parser integration with a runtime system would be nice, a lot of the DSL-happy people just want to be able to capitalise on syntactic and semantic side-effects of their favourite language, calling for conveniences that a language designer should resist. (And they could all go off and play with Lisp anyway, right?)

It has undoubtedly been said that good designers know that leaving features out can be more important than putting features in. Looking at that Ruby slideshow, one can but wonder whether, after adding things already in Python but then overshooting to include all sorts of other one-off conveniences, Matz will be as conservative as GvR supposedly is today.

By: Cezar

Cezar — Thu, 29 Mar 2007 03:48:28 +0000

I find it a good thing that Python is not going to change much. When looking to do real work, I look for a few things. The first and foremost is readability, I don’t want to spend all day reverse engineering someone else’s code. After I move on, my replacement should be able to read my code. While magic is fun, try being the one who has to decipher it.