How to become a proficient Python programmer
blog.dispatched.ch
blog.dispatched.ch
Idioms are indeed important, but TDD won't improve your proficiency (except in that you'll learn a testing framework and be writing more code). Neither will attempting to adhere to a "Functional" style when you've not already learned the important lessons of FP from an actually FP-oriented language. Coding guidelines? Be consistent. Discover the guidelines of the code you're working on and use that. That has nothing to do with python or proficiency.
Performance is probably the last part of proficiency that one has to learn, and arguably the most difficult. A lot of other skills have to come together before one can count "performance" as a point of their proficiency in a language, and most of those skills are really quite language-agnostic. Oh, and if you're focusing on performance but you don't yet know how to properly profile and benchmark, you're not proficient.
If you really want to become a proficient Python programmer (or any other language) try this:
1. Find the documentation
2. Learn how to navigate, search, and read the documentation
3. Find a skilled mentor or some community resource (IRC, mailing-list, website forums, etc)
4. Ask what is considered good code. Read it. Study it.
5. Create a project - little ones, like sorting your MP3 collection or mining your email for statistics.
6. Try to use the lessons learned in the code you've studied when writing your project.
7. lather, rinse, repeat. Chances are, you're not proficient until you're doing it for several hours every day - and even then, you're probably not really "proficient"... yet.
/rant off
Any suggestions?
Another strategy is to skim the source code for your dependencies, without worrying about whether the code is good or not. You'll still get a good idea of the scope of the problem the library is solving, and whether it looks like the authors knew what they were doing. You'll have a crude map of the code in your code, useful when you run into bugs later.
The number of classes and interfaces in Twisted is a bit of a stumbling block, but I've never come across anyone with a better, more "Pythonic" way to express the same ideas or provide the same flexibility. The ABCs introduced in modern Python are inferior to the interfaces Twisted uses, and the most recent PEP I've seen for Futures is but a pale shadow of Twisted's Deferreds.
1. Reading a library you are using, even if it's a standard library like smtplib is fun and makes you smarter and your code better.
2. Python is such a joy to read, unlike some other languages.
3. Some libraries (I'll mention smtplib again) are such a mess, and yet reasons 1 + 2 still make it easy for you to learn from them, while also learning what not to do.
> TDD won't improve your proficiency
Disagree. Knowing that you need to test your code, and going back and testing your old code gives you a real feel for how to split out code, how to write interfaces, and how to organize your objects - at least in Perl and JS, which are the languages I'm proficient in.
If you know that your routine that'll be accepting an object might be accepting a mocked up object, you're more likely to resist the temptation to get all molesty with its private methods; if you know you'll spend time trying to drive up the code coverage, you'll be more inclined to abstract and not repeat yourself.
Random plug for a book that made me a waaaay better Perl programmer: http://perlmedic.com/
Also, yeah... I should take-back the TDD comment, or at least re-word it. I just get a little knee-jerky when I hear the word TDD mindlessly spouted out as a panacea. TDD works best when combined with other best-practices, and of course, learning how to write tests that help and don't become yet another ball of mud is an art in itself!
And its HN discussion: http://news.ycombinator.com/item?id=1890494
for example, about 6 months ago i rewrote someone's C code into python. this was numerical code for processing astronomy observations and comparing them with numerical models. i reduced the running time from about 24 hours to 15 minutes. that wasn't because python is faster than C (it's not), or because i used python specific idioms (i did, but they made no real difference). it's because i replaced O(n^2) algorithms with linear ones.
the lesson being: language is not important.
Runtimes of languages still matter. Same algorithm implementation may run significantly slower in some languages than others (and this is not necessarily the language issue but interpreter/compiler issue).
Here are two good Stack Overflow threads:
http://stackoverflow.com/questions/1092675/what-ruby-knowled...
http://stackoverflow.com/questions/2569501/what-ruby-and-rai...
I'm the author of the article, so thanks a bunch for the praise! I have switched to being a professional Rails programmer four months ago and left the last 2.5 years of daily Python programming behind me. This post is more a by-product of a mail that I answered.
Maybe I'll sometimes get around and write a pendant for Ruby. To get you started - for my new job I had to read these seven books:
* Agile Web development by Sam Ruby
* CSS, the definitve guide
* Head first software development
* jQuery in Action
* pragmatic git
* The Rails Way by Obie Fernandez
* The Rspec Book
Granted, this is less Ruby then Rails specific.
Best to you on your path,
Alain
Thanks you so much for the tips and the list. This is why I love Hacker News.
it is about giving you a tool to deeply understand your own problem domain
However that's not what TDD actually does: http://devgrind.com/2007/04/25/how-to-not-solve-a-sudoku/
Btw, Peter Norvig himself has good things to say on TDD[1]. So using him in an argument against TDD whilst he's using tests is kinda self defeating.
1. http://gigamonkeys.wordpress.com/2009/10/05/coders-unit-test...
Then he isn't doing TDD at all is he?
"Test-driven development (TDD) is a software development process that relies on the repetition of a very short development cycle: first the developer writes a failing automated test case that defines a desired improvement or new function, then produces code to pass that test ..." [Wikipedia]
Using tests after you write your code, to test functionalty how it needs to be improved has been a common practice from the earliest days of sw dev. In other words, "the importance of tests" has never been in dispute. TDD is not so universally accepted.
Testing is beneficial, sure. But the claim that TDD ,sometimes evangelized as "not a testing method, it is a design method" (roll eyes), actually helps in "deeply understanding your problem domain" needs more validation.
Due Disclosure: I wrote the original blog post(EDIT: the blog post comparing the two Sudoku attempts not the OP for this HN discussion) . I personally think TDD is overhyped consultantware snake oil, but that is just my personal opinion. If it works for you, more power to you.
Btw, Peter Norvig himself has good things to say on TDD[1]. So using him in an argument against TDD whilst he's using tests is kinda self defeating.
1. http://gigamonkeys.wordpress.com/2009/10/05/coders-unit-test...
(Tho' I do think Ron Jeffries is a snake-oil peddler).
I think in general that Ruby has one of the better communities out there (I'm a rubyist myself) but all this noise about TDD and cute DSLs (RSpec, Cucumber...) hasn't made a material difference in Ruby code quality compared to any other language.