Why I Hate Python (Or Any Dynamic Language, Really)


A project I'm currently doing for work involves a custom server-side component. Ordinarily I would reach straight for D and Vibe.d. (In fact, D and Vibe.d are already making development go very smoothly on a light-weight "not-a-blog" system I'm developing...at least in the rare cases I can actually spend any time on it...) Unfortunately, this project came with an unwavering pointy-hair decree that I must use either PHP or Python. (BTW, If you think you know who it was, you're probably wrong.)

Naturally I didn't hesitate to choose "Anything but PHP".

But, it's worth noting the big reason behind the "PHP or Python" limit was the usual false-dichotomy bullcrap about "getting shit done" that's frequently used by dynamic fans. About the only argument that made any sense at all was that a lot of people know PHP and Python and could pick up the codebase later (It's always nice to know you're coding your way towards expendability.)

Whatever, fine. Python it is.

So, long story short, I found this page which pointed me to some relevant libs, and I figured I'd try Flask first.

I went to download Flask (BTW, if I see one more language or lib claiming on its homepage to be "fun!", I'm gonna kill someone). But Flask showed nothing but a Linux release. WTF? I thought Python was supposed to be at least vaguely cross-platform? Searched around and found some obscure corner of the net that happened to actually explain how to install it on Windows. Ok, so I installed the prerequisite PIP, I do the "pip install flask"...and the system couldn't find PIP. More digging...Oh great, back when I had to put the Python directory into my PATH manually, despite having used an actual Python installer, I was apparently supposed to also know to add the 'Scripts' subdirectory. Clear as mud.

Fine, done. Try again...And I get some big useless "Traceback" vomit. (I swear, when trying to use Python "software", I get more "Traceback" bullshit than actual "Ok boss! Done!".) Buried in that garbage is some line of sourcecode involving a string literal mentioning something about needing a "PyFlakes". Ehh, what the hell, let's just try "pip install pyflakes". Nah, no dice, just more barf.

Meh, whatever, fuck Flask for now, let's try Bottle. Apparently it's supposed to be low-bullshit anyway, so sounds good. Wow, this one actually installs. And the "hello world" source on their website works! Hot damn, we're in business.

So I modify the hello world. And I look up how to change the content type, because I know I'm going to need to spit out binary data on this project. And...uhhh...the content type didn't fucking change. It's still spitting out 'text/html'. Errghh...So I blow nearly half an hour digging through the docs for something I may have missed, go back over my code...nothing. Try to find someplace to ask, find a buried link to a mailing list (I hate mailing lists), and start writing for help. Then, by pure chance, I happen to notice it:

from bottle import route, run, request, response @route('/foo') def index(): request.content_type = 'application/octet-stream' return 'Blah' run(host='localhost', port=8181)

Do you see the problem? I mean, aside from the flawed idea of indent-scoping and the goofy underscore_naming_convention. Yea...Obviously "content_type" is a property of the response, not the request. Of course, Python couldn't have told me 'request' didn't have a member 'content_type' and ended the matter right then and there. No, it had to string me along, doing anything except let me know something was wrong.

If this is a dynamic coder's idea of "getting shit done", then they can blow me.

Oh, sure, you never make stupid mistakes like that, right? There's a term for programmers who don't make dumb mistakes: Goddamned LIARS.

After fixing it to "response" we can still have some more fun with this:

response.content_typePOOP_SHITS_FUCK = 'application/octet-stream'

Still runs! Wrongly! Even though I'm damn certain the Bottle developers didn't put any "POOP_SHITS_FUCK" into their API.

Why the fuck should any of that even compile? That's what I hate about these dynamic turds, they happily let you do completely nonsensical shit, just for the fuck it. "Oh, but this means you can stick random extra garbage into any object you want!" Gee, great. Just what I've never wanted.

Here's another wonderful feature I never wanted:

from datetime import date ... response.content_type = date.today()

Holy hell, my content type is a fucking day and Python happily spits out the resulting gibberish. Gee, it's not a bug, it's a feature! (Hmm, happily doing the non-sensical? No wonder it's named after Monty Python[1].) "But what if you wanted to..." Lemme just stop you there. The day I want to do something like that is the day I want my head bashed in.

"But it's just doing what you clearly told it to!" Great, I see we're already forgetting something about "programmers who don't make mistakes". Now, I don't like software that second-guesses me any more than anyone else, but when I fuck up (as we all do) and tell my computer to do something that doesn't even make sense, I expect it to tell me so, not use it as an excuse to run around fucking shit up. If I hand you a gun and say "Shoot the fribstitch", I expect you to say "What the fuck is a fribstitch?", not start blasting away hoping to hit it.

So let's recap: Dynamic typing helps you "get shit done" by giving you the freedom to do useless nonsensical things, write to properties that don't even fucking exist, and generally do things that are either outright bugs or are convoluted enough to invite bugs. Then, instead of letting the compiler eliminate entire classes of these bugs, you're encouraged to take the time to write extra unittests to catch some of them instead of, you know, just "getting shit done".

If I'm gonna "get shit done", then I expect my language to help me avoid bugs, not help me make them.

[1] I am indeed a fan of Python the Monty. Just not Python the Dynamic.

40 comments for "Why I Hate Python (Or Any Dynamic Language, Really)"

  1. (Guest) isomorph
    2012-10-21 00:25

    LMFHO i couldn't have said it better. learning JS right now--i know, just shoot me--because i got a web app to build and there is nothing else running client side. i can't believe that we've been stuck with this garbage for more than 15 years now. i remember when netscape was pushing it. a took 1 look and my stomach heaved. that was it for me. had i not had c/c++, and later java, it would have been the monastery somewhere far away from the madness. server-side there seems to be now an inordinate amount of juvenile enthusiasm for something called node.js. thank god there is scala, akka, and typesafe. didn't know about D until yesterday. problem is, there are all these libraries available now to speed up development server-side with dynamic languages. don't know what D has, but if one has to reinvent the wheel... well you get my point.

  2. 2012-10-21 13:44

    Yea, funny thing about JS and Python: As horrible as JS is, even *it* at least has optional static type annotations in later versions of the ECMAScript spec. Python doesn't even have that.

    The big server-side lib for D is Vibe.d < http://vibed.org >. I use it and I highly recommend it. It's sorta like node.js except it doesn't suck.

    D also has this set of libs: < https://github.com/adamdruppe/misc-stuff-including-D-programming-language-web-stuff > It's not very well documented, but it has a lot of web-related stuff, including a D server-side HTML DOM, which is very cool. I use it instead of the HTML templating engine that's built into Vibe.d.

  3. (Guest) isomorph
    2012-11-01 20:37

    thanks. greatly appreciated.

  4. (Guest) isomorph
    2012-11-10 02:31

    hey! i was wrong regarding being stuck with JS. just stumbled today onto Google's new language whose goal is to get rid of JS, at least on the front -end. the language name is: DART. at first glance, looks like a better tool than JS.

  5. (Guest) sudo !!
    2013-02-20 11:53

    The only dynamic language I tolerate is Java. Others? Go to hell.

  6. (Guest) guest
    2013-02-23 12:12

    Is Java a dynamic language? I don't think so...

    At least that's what I would have said if you asked me off-hand.

    But reading wikipedia it looks like Java *could* fit into the definition of a statically typed dynamic language.

    Learned something new.

  7. (Guest) Milan
    2013-07-09 21:10

    I was debugging my WordPress plugin for the past 26 hours and naturally, after no progress whatsoever, I have decided to type "fuck dynamic languages" into Google, and that is how I get here.

    Thank you for the article.

  8. (Guest) guest
    2014-01-06 08:46

    Python has its own cons, but mistakes you've made just make me laugh. You are shitty programmer and thats it.

  9. 2014-01-11 20:06

    @guest: For the third time, since the first two mentions were *cough* mistakenly overlooked:

    "There's a term for programmers who don't make dumb mistakes: Goddamned LIARS."

    What separates the good programmers from the rest, as with *any* form of engineering, is the ability and willingness to recognize that inevitable fallibility and plan for it, minimizing its impact.

    All people fuck up. But the true talent are those smart enough to have already set up failsafes.

  10. (Guest) guest
    2014-01-22 13:40

    I always joke and say that you can judge a programmer by what they blame:
    The novice blames the OS;
    The intermediate blames the middleware (<--- you);
    The craftsman blames himself.

    On this whole page I found only 1 truth, and that was your sub-title:
    "Inflammatory rants"...

    All of the best mate.

  11. 2014-01-28 03:12

    That's an old saying and people constantly interpret it wrong.

    The craftsman blames himself *because* it's his responsibility to pick the right tool (and *then* use it correctly).

    Being a good craftsman is NOT about grabbing any random junk and expecting to do an adequate job with something ill-suited for the task. I don't understand where so many people keep getting such a bizarrely twisted and nonsensical interpretation.

    If you drive a nail with a hammer and fail: Your fault for using the hammer wrong.

    If you drive a nail with a screwdriver and fail: Your fault, but only because you made a moronic choice by selecting the screwdriver in the first place.

    It doesn't matter one bit whether it's the screwdriver's fault or the craftsman's fault for picking the screwdriver. That distinction, as well as the entire blame-game it comes with, is semantic bullshittery. Doesn't matter what's to blame; what's important is that using a screwdriver to drive a nail is moronic.

    ...Just like using a dynamic language to write non-trivial software. And no amount of trite proverbs, let alone misinterpreted ones, can change that.

  12. (Guest) The_Frenchman
    2014-05-14 10:35

    Thank you so much for the good laugh.

    After spending an entire afternoon trying to figure out why Mr. Python kept evaluating to True the critical "If x < y:" line in my code, regardless of x's and y's values of course, I finally nailed it: x was a good old float but y was a.. string. My bad, ok, i'll admit to forgetting the appropriate type casting. I suppose strings are implictly valued to 0 or something so comparing numbers with them can "get the fuck done"...

    Boiling with wrath (unlegitimate I'm sure, easy, easy Python Ayatollahs), I decided for some reason to try and soothe it down by googling "Python is ****" which brought me to your fribstitch article... Hilarious and thoughtful.

  13. (Guest) guest
    2014-05-14 16:23

    Google fuck Python. Was not disappointed.

  14. (Guest) guest
    2014-05-14 16:27

    Oh and I almost forget : http://benchmarksgame.alioth.debian.org/u32/benchmark.php?test=all&lang=php&lang2=python3&data=u32

    Yes... they both are shitty languages but Python is even much more slower than PHP >_<

  15. (Guest) FUCKPYTHON
    2014-06-09 03:22

    Python was made by crazy monkeys ...

  16. (Guest) alex.k
    2014-06-17 13:35

    I feel your pain!
    You've just described my Python experience.

  17. (Guest) guest
    2014-06-24 16:16

    But look at the bright side
    While it allows you to write bull shit code, it will force you
    to indent it correctly.

  18. 2014-07-15 10:21

    > While it allows you to write bull shit code, it will force you
    to indent it correctly.

    Hah! I like that! :)

  19. (Guest) Kalrish
    2014-08-27 16:25

    I found the article as others have mentioned, and it likewise made me smile :). Perhaps I am too dumb or haven't really given Python an opportunity, but I too feel the same. Programming languages, in the first place, serve to offer concepts over the raw stream of bytes; what's the purpose of a language without strict types, that allows for such nonsensical, non-obvious behaviour? With SCons I already had some problems, and the same happened for some simple tools that I decided to implement with Python that ended up not being so simple. It seems as if what you get with the 'dynamic' part is what you later pay guessing how some trivial things are happening. For now, I'll keep with C++ :).

    (Again, this is just my current feeling. I realize that I might not have had a real experience with Python, and that it could change.)

  20. (Guest) guest
    2014-09-14 10:23

    Python was intended as an educational language and so on every opportunity it gets it tries to make you limp along to a new enlightmentment.
    Getting shit down isn't even remotely in the scope and nor is getting you just the info you need to fix your error.

    This is surely awesome if all you do is sit at some university getting knowledge stuffed up your ass.

    In other contexts, it can be a bit bewildering why you're told there was an unexpected end of the line, and that that occured while it was scanning a literal that was apparently a string instead of simply saying missing "/' in line 1510.

  21. (Guest) guest
    2014-12-04 01:34

    Yes, this, thank you.

    Fuck Python.

    I don't care for your inflammatory 'anything but PHP' nonsense though.

  22. (Guest) Steve
    2015-03-01 18:08

    When I heard someone said "Dynamic language could have beated statical language", I immediately searched "Fuck dynamic language", the "language" doesn't ends with a 's', is because I hated it from my bones. Nevertheless it ends up leading me to here, and expressed half of my feeling towards the "dynamic language", but he doesn't mentioned that so many "popular tools" are based in "dynamic language": Angular, Web.py, Ruby in rails, node.js and sort, where the common point among them is, they are totally non-sense bullcrap that you have to invest in a lot of time to debug just for one god-damned line of mistake, however the "programmer of nowadays" use them, and often leads to bloat-wares that you would screw you life by reading their documents oftenly, and I finally give up learning the "dynamic language", people who use "dynamic language" is the same who use BASIC and need to be corrected immediately, or otherwisr they will kill the normal and turn non-sense into new "normal".

    *Forgive my swear, but BASIC is the bullshit-est even far ahead of dynamic languages, people who uses BASIC as their primary programming language don't even know what the fuck is structure but know what the fuck is goto and spagetti code.

  23. (Guest) guest
    2015-04-16 18:58

    After a day of frustrating python coding, this is exactly what I needed to read for a break. Good to know I'm not alone.

  24. (Guest) guest
    2015-05-08 07:40

    Just goes to show, there are gotchas in every language.

    Frankly, this example is bordering on ridiculous. Reading between the lines, it seems like frustration bourne of dealing with unfamiliar territory.

    Maybe you'd be happier in .NET.

  25. (Guest) guest
    2015-05-08 08:11

    jak się nie umie to się nie potrafi

  26. (Guest) grzegorz brzęczyszczykiewicz
    2015-05-08 08:12

    pierdolenie - jak się nie umie to się nie potrafi

  27. (Guest) guest
    2015-05-08 16:21

    You are doing it wrong

  28. (Guest) guest
    2015-05-08 21:34

    Clearly this is the worst reasoning I've ever seen against Python. He/She's annoyed because the language doesn't magically opine intent and suggest an alternative (even when request DOES have a content type property)?? That it doesn't run cross platform?? I'm curious as to what language the author does feel to be ideal across the full spectrum of preordained cynicism covered here.

    What's ironic is that ultimately Python allowed him/her to prototype the intended functionality and test such within minutes (and with a bit more time, provides one of the most powerful unit testing frameworks to idiot proof his/her code from such future mistakes). Love to see how fast that goes in a large scale C++ app which can take hours to build before discovering such a mistake - and failing - requiring at least another few hours to try again, and that's before even attempting a single functional test. I won't even mention java as the callouses from all the typing would speak for themselves.

    There are clearly some sacrifices and other shortcomings with Python but I'd anything in THIS article is enough to convince someone to stay away from Python, clearly you are inexperienced enough to be easily swayed by passionate, superficial and illogical hyperbole

  29. 2015-05-09 10:27

    > He/She's annoyed because the language doesn't magically opine intent and suggest an alternative

    "opine intent" != "tell you when something actually went wrong". As explained in the post, I'm only advocating the latter.

    > even when request DOES have a content type property

    Shouldn't be public-writable anyway. But even if it wasn't there, it still would have done the same "Well, this property being accessed doesn't exist in the API, but let's just pretend nothing went wrong and keep chugging along anyway".

    > ... C++ ... Java ...

    Equating non-dynamic language with "C++ or Java" is hyperbole as well. That would be just as nonsensical as if I had said "dynamic languages are all terrible *because PHP*".

    See, it works both ways. You can't argue against an entire class of languages by cherry-picking the two absolute worst examples (C++ and Java).

    BTW, this page shows what it looks like in D:
    http://vibed.org/

    That's some pretty damn quick prototyping, too ;) And note that adding "res.contentType = Clock.currTime" or "res.contentTypePROP_DOES_NOT_EXIST = "whatever"" results in immediate compiler feedback of the mistake. I guess it does opine intent!

  30. (Guest) tommed
    2015-05-10 02:30

    Nice rant! I'm no fan of python either, no interfaces, not even duck-typing, but the mistakes you made would have been covered by unit tests. Had you made the mistake inside the content type string, all compilers would have missed it. Therefore having an accompanying unit test would ensure that 1. You've set the correct property and 2. The value you've set it to is also correct.

  31. 2015-05-10 09:58

    @tommed: Unittests are indeed essential for modern software development, even in static languages.

    In this case though, it wouldn't have helped: Detecting that the property wasn't getting set correctly wasn't the problem. I immediately detected that as soon as I ran it and saw it spitting out "text/html" instead of "application/octet-stream". The problem is the compiler responding to things that shouldn't be valid code in the first place by ignoring them rather than flagging them.

    I thought Python did do duck typing though?

  32. (Guest) gigi
    2015-05-11 11:55

    Thank you man, I'm forced to use this 'funny' thing that modern hippies love so much, and I hate it, I hate it with all the cells of my body. I'll come back and re-read this every time the through 'fuck this shit' start building up in my mind, to reset it and keep working on this fucking shit. Best regards.

  33. (Guest) guest
    2015-05-11 12:56

    Garbage in.
    Garbage out.

  34. 2015-05-11 17:00

    Weak/Dynamic: Garbage in, Garbage out.

    Static: Garbage in, Meaningful diagnostic message out.

  35. (Guest) guest
    2015-05-16 10:51

    I agree. I don't see the advantage of using Python unless you are doing very simple stuff. I rather use C# or Java for the same reasons you provided. Sometimes you don't have time to write unit tests and prefer to write code towards the product.

  36. (Guest) guest
    2015-06-09 09:47

    I so hate dynamic languages. Every time I work with one, I end up with implementing a complete type system in my tests. I am at lost how people "get shit done" with this crap.

    And the programmers must be completely retarded or something, why do you need the freedom to write error prone code? Is it because they don't really understand what they write?

  37. (Guest) guest
    2015-06-12 16:06

    The correct way to set mimetype: http://stackoverflow.com/a/11774026/2534876G

  38. (Guest) guest
    2015-06-25 13:26

    What you mentioned are weaknesses of Python (and Ruby) but not weaknesses of dynamic typing. I can also rant about how crap static typing is by referencing Java, C#'s or C++'s type systems but I'll immediately get a response have some Haskell or Ocaml guy telling me not to judge static typing by the worst examples.

    Let me just give you an example of why Java/C#/C++'s type systems are bad: I have simple task divide to numbers and get a decimal fraction as a result, in a dynamic language that is no problem since variables carry type information at runtime they will be coerced into the appropriate type thus you just divide the numbers. In Java however If I just divide the numbers e.g. 'a/b' I will get the result I desire IF and ONLY IF a and b are both floats/doubles, however if they happen to be ints the result will be an integer even if what I am doing is assigning the result to a float/double I will not get some useful diagnostic type error message, nor will I even get a runtime crash, instead the result will be implicitly truncated and a bug may not even be visible until MUCH later. In order to make this work I have to do put casts everywhere (double)(a)/(double)(b) yea other than not being readable, you're performing a cast isn't that going against the purpose of static typing anyway.

    Either way what you wrote are weaknesses of the Python programming language, in other dynamic languages such as Lisp don't have the aforementioned problems.

  39. (Guest) guest
    2015-07-17 01:25

    I don't really have a dog in this fight. I'm an independent contractor and I end up working with whatever is required regardless of my opinions. I have to say though every time I have to work with a non-trivial set of Python code it's always a steaming pile of shit. I don't know that the problem is the language so much as it is the culture of Python programmers. It's like because they're not forced to apply rigor then they are somehow excused from it. I'd say 60-70% of the Python code I've had to fix over the years was written by non-developers. Various kinds of engineers, academics, system administrators, etc. Basically otherwise intelligent people who just don't know what the fuck they're doing when it comes to developing software but don't yet realize it. I think Python is ok for building test drivers and sysadmin scripts and things like that but for big boy work I reach for C++, Scala, Java if I must and C# for any .NET or Mono work.

  40. (Guest) guest
    2015-08-21 04:56

    Python is Elegant language. Stop bashing Python you don't know true power of Python

Leave a comment

Captcha