The "IT DOESN'T WORK!" thread

[Cosmologicon]

Python doesn't have true global variables, only module-level variables. "Globals" defined in one module aren't visible in another.

If module m defines a "global" v (just with something like v=10 outside of a function), then other modules that import m can refer to it via m.v, but you need the module qualification just like anything else. (I think that the generally-ill-advised import * from m would make v available unqualified.)

It's from m import *. And the perfectly fine from m import v will also allow you to refer to v unqualified.

[Windowlicker]

Working on a Haskell implementation of code war.. it's mostly done. VM is working, parser is working, a GUI is well on the way.. the only place I'm having trouble is reading warriors from files and adding them to the core. I'm getting an infinite loop, and I think it's in here:

Code: Select all

makeState :: IO [Warrior] -> IO System -> IO System
makeState inp state = do
    i <- inp
    st <- state
    if length i == 0 then
        return st
    else
        makeState (return (tail i)) $ genState (head i) st

I've tried a few subtle variations of this, but I still can't get it to work.

What I'd like to do is

Spoiler:

Code: Select all

makeState :: IO [Warrior] -> IO System -> IO System
makeState [] state = state
makeState (x:xs) state = do
    y <- x
    st <- state
    makeState xs $ genState y st

but.. you know, there's that whole can't pattern-match with monads thing.

genState is genState :: Warrior -> System -> IO System. It works when used outside this context, so I don't think it's where the error lies. It takes the warrior and system, then adds the warrior to a random location and returns the resulting system. I.. cannot figure out why this doesn't work.

[phlip]

Sounds like a job for a fold...

Code: Select all

makeState inp state = do
    i <- inp
    st <- state
    foldlM (flip genState) st i

That said, I'm not sure what's wrong with your code... are you sure it's infinite-looping? How big is the list you're passing for inp? Could be it's just taking a while to run... you can move up from O(n²) to O(n) by replacing "length i == 0" with "null i" (assuming genState is O(1))...

[Windowlicker]

I thought a fold might have been more.. elegant, but I couldn't work out if it was exactly what I wanted.
I'm assuming it's an infinite loop. Compiling the code with ghc and running it gives the output "Main: <<loop>>". I couldn't find any information about it, so I could be wrong with my assumption.
The lists are only 2/3 long at the moment, so it shouldn't be taking too long to process them.
I've tried the fold, and it's producing the same output. Ah, I don't know.

It's difficult to tell, 'cause we have a function called loop also. The test code that worked was:

Code: Select all

main :: IO ()
main = do
    st <- genState dwarf init_state
    st' <- genState dwarf1 st
    loop st' 0

Dwarf and dwarf1 are test warriors, hard-coded into another module.. they're what I'm trying to load from files.
The main loop that doesn't work is:

Code: Select all

main :: IO ()
main = do
    sys <- loadSystem
    loop sys 0

Where loadSystem :: IO System.

[Jof16's]

Python doesn't have true global variables, only module-level variables. "Globals" defined in one module aren't visible in another.

If module m defines a "global" v (just with something like v=10 outside of a function), then other modules that import m can refer to it via m.v, but you need the module qualification just like anything else. (I think that the generally-ill-advised import * from m would make v available unqualified.)

It's from m import *. And the perfectly fine from m import v will also allow you to refer to v unqualified.

Okay, so I had to rewrite the entire thing to allow for the variables to be accessible from the module.
Now I have a question. Several of the variables need to be changed when a certain button is pushed. However, using 'variable = 10' inside the functions that these buttons call doesn't seem to be working. Am I doing something wrong?

And thank y'all for the help so far.

[Windowlicker]

If you're trying to access a global variable (let's say x), you'll need to include the line "global x" inside the function definition before you can change it. I believe you can read them without this call.
I haven't used this with variables from another module, but based on what Evan's said it might be something like "global m.v".
Python namespacing/scope.. takes a bit of getting used to.

[Jof16's]

Right now, I'm using the ' from x import * ' method. I am using 'global v' inside the function, but it isn't changing the variable.
and yes, you can read the variables without the global declaration (is that the right word?)

Yes, it does.

[Cosmologicon]

If you're trying to access a global variable (let's say x), you'll need to include the line "global x" inside the function definition before you can change it. I believe you can read them without this call.
I haven't used this with variables from another module, but based on what Evan's said it might be something like "global m.v".
Python namespacing/scope.. takes a bit of getting used to.

Yeah it does take some getting used to, but at least it's consistent. If you're referring to a variable as m.v, you don't need to specify that it's a global. As long as you never change the value of m within the function (m = .... or something like that) then m is automatically a global, referring to the same thing it refers to outside the function. And since m is a global referring to the module, m.v refers to the property v of the module m. So these work just fine:

Code: Select all

import m
def setv(value):
    m.v = value

Code: Select all

class X: pass
m = X()
def setv(value):
    m.v = value

But for these you need to specify global:

Code: Select all

from m import v
def setv(value):
    global v
    v = value

Code: Select all

v = 0
def setv(value):
    global v
    v = value

I suspect, however, that Jof16 is doing it "wrong". The global keyword is probably not the "right" way to handle this case (although I use it all the time when it's necessary). A minimal code example might help.

[Windowlicker]

The third example was what I was aiming for, just with.. well, mistakes in. The first way is the way I'd do it myself, but for some reason I've recently become very into qualifying imports in every language I'm using..

[Jof16's]

Here's some of the code:

Code: Select all


def buttonSword():
    global weaponDamage
    global weaponAmmo

    weaponDamage = 5
    text.insert(END, "You have chosen a sword.\n")
    text.insert(END, "It can do up to " + str(weaponDamage) + " points of damage.\n")
    text.insert(END, "To change your weapon, press one of the weapon buttons\nbelow.\n\n")
    weaponAmmo = ammo(weaponDamage, text)


weaponDamage has a value of 0 in the module that contains all the global variables.
I keep getting the error at this line:

Code: Select all

zDamage = randint(1, weaponDamage)


There are also several others that won't change, but they won't throw errors because it just refers back to the original value each time through.

[Cosmologicon]

The third example was what I was aiming for, just with.. well, mistakes in. The first way is the way I'd do it myself, but for some reason I've recently become very into qualifying imports in every language I'm using..

Let's be clear, the first and third examples are not quite doing the same thing. If you use "import m" and then set m.v, you're actually changing the value of the property v in module m. If you use "from m import v", then you're creating an identifier v that refers to the same object as property v of module m, but then if you reassign v, you're no longer referring to that same object. The value of m.v will be unchanged. Here's an example. Say my m.py file consists just of "v = 0":

Code: Select all

>>> import m
>>> m.v = 1
>>> import m
>>> m.v
1

Code: Select all

>>> from m import v
>>> v = 1
>>> from m import v
>>> v
0

Of course, if you never change the value of v, then it's not an issue.

[Cosmologicon]

Here's some of the code:

Code: Select all


def buttonSword():
    global weaponDamage
    global weaponAmmo

    weaponDamage = 5
    text.insert(END, "You have chosen a sword.\n")
    text.insert(END, "It can do up to " + str(weaponDamage) + " points of damage.\n")
    text.insert(END, "To change your weapon, press one of the weapon buttons\nbelow.\n\n")
    weaponAmmo = ammo(weaponDamage, text)


weaponDamage has a value of 0 in the module that contains all the global variables.

How are you initially defining weaponDamage within the module that contains buttonSword? Are you saying "from globalvars import weaponDamage"? If so, then take a look at my last post, this function of yours is not actually setting the variable within that module. I think it would be better if you just say "import globalvars" and then within buttonSword set globalvars.weaponDamage.

[Jof16's]

I'm using ' from Game_config import * '

Okay. I'll go try that.

Thank you.


EDIT: Yay! It works!
Thank y'all for the help.
Also, if you would please check out my thread. I've still got some problems that need solving. Thanks in advance.

[Windowlicker]

The third example was what I was aiming for, just with.. well, mistakes in. The first way is the way I'd do it myself, but for some reason I've recently become very into qualifying imports in every language I'm using..

Let's be clear, the first and third examples are not quite doing the same thing. If you use "import m" and then set m.v, you're actually changing the value of the property v in module m. If you use "from m import v", then you're creating an identifier v that refers to the same object as property v of module m, but then if you reassign v, you're no longer referring to that same object. The value of m.v will be unchanged. Here's an example. Say my m.py file consists just of "v = 0"

Aaah, of course. This is useful information.. Despite the fact that I have been and currently am working with Python on a few projects, I've not come across the need to edit variables in another module. But, the more you know, eh?

As for my Haskell issue, I got that fixed: it turns out the error was a few lines later, I'd declared "let war = war ++ [listWarriors]" or something like that to build the list. Problem solved, easy change to "let war' = war ++ [listWarriors]". Ah, haskell.

[headprogrammingczar]

If you REALLY don't want to name variables "war'", you can do

Code: Select all

x # f = f x

war ++ [listWarriors] $ \war ->
...


You would essentially be doing a CPS-transform of the code around that 'let'.

[EvanED]

How can I make diff ignore all whitespace differences, even across lines?

edit Here's my "solution": diff --ignore-all-space <(cat $file1 tr -d "\n") <(cat $file2 tr -d "\n").

[not baby Newt]

The googles suggests you shouldn't need to go that far. But I haven't tried it out and you already use the option mentioned.

The `-w' and `--ignore-all-space' options are stronger than `-b'. They ignore difference even if one file has white space where the other file has none. "White space" characters include tab, newline, vertical tab, form feed, carriage return, and space; some locales may define additional characters to be white space.

[EvanED]

Yeah, that doesn't work, at least on my (old, RHEL5-supplied) version of coreutils:

Code: Select all

$ diff -w --ignore-all-space <(echo "ab") <(echo "a\nb")
1c1,2
< ab
---
> a
> b

OTOH, my man page says

Code: Select all

       --ignore-all-space
              Ignore white space when comparing lines.

which is actually correct, I guess.

[Aschey]

I just installed Python (for Mac), typed in a simple command, print 2 + 2, saved the file, and tried to run it by dragging the file into the Python Launcher. No matter what I try, I keep getting a syntax error in Terminal. I checked my syntax, tried other basic commands, and keep getting the same error. Anyone know what's up? Thanks.

[TheChewanater]

"print 2 + 2" is not valid in newer versions of python, so you may need to write it as a function - "print(2 + 2)". Can you post the error you're getting, and the version of Python you're using?

[Aschey]

I'm not using the newest version, as it said it wasn't compatible with some things yet. But I fixed the issue, apparently I had to open a new window after opening the initial program. Not sure why that made a difference, but whatever. Thank you for the reply though.

[Jplus]

I just installed Python (for Mac), typed in a simple command, print 2 + 2, saved the file, and tried to run it by dragging the file into the Python Launcher. No matter what I try, I keep getting a syntax error in Terminal. I checked my syntax, tried other basic commands, and keep getting the same error. Anyone know what's up? Thanks.

As an aside, did you know Python is already installed on every mac by default?

[PerchloricAcid]

This isn't really a ''it doesn't work'' kind of problem/question, but I haven't found a thread dedicated to asking BASIC "how to" questions. If such a topic exists, please point it to me. My question isn't really worth opening a whole new topic.

So:
i have a .txt file with some integers. I'd like to write a program in C++ which would take every Nth integer from the file (and save it in another file). How?

[RoadieRich]

This isn't really a ''it doesn't work'' kind of problem/question, but I haven't found a thread dedicated to asking BASIC "how to" questions. If such a topic exists, please point it to me. My question isn't really worth opening a whole new topic.

So:
i have a .txt file with some integers. I'd like to write a program in C++ which would take every Nth integer from the file (and save it in another file). How?

It's not a good idea to capitalise basic like that in a programming forum... it means something very different if you do that.

As to writing your program, you open a text editor, and type in the code.

This sounds suspiciously like a homework problem, for which we won't write the code for you, but we will help you, if you're willing to help yourself. To start with, show us what you can do.

[PerchloricAcid]

I didn't even realize I capitalized it, however, I did it because I wanted to emphasize that my question wasn't worth a new topic. I am not the type of person that comes and asks other people to complete their homework or any kind of work for them. I just want some hints.
And, in case you saw my reply before the edit, I apologize for it being written in an unpleasant manner.

As to what I know:
I believe I could do this by using a for loop, where the increment would not be 1, but N. Perhaps I'd need to check whether I have reached the end of the file, so it might be better to use a while statement, although I wouldn't know how to check whether I have reached the end of the file (nor is it necessary).
I know how I'd save the taken values to another file, but I don't know how exactly would I "pick out" the values I want. I believe I might be needing "ifstream".
Of course, I'd open the file I'm reading from at first, I'd have a line which checks whether it really is opened, and close it in the end.

[Xanthir]

If you know how to do a for loop, and you know how to read a line from a file and write out to a different file, then the piece you're missing is the modulus operator "%". When you do "a % b", it returns the remainder of "a / b". This result cycles through the integers between 0 and b, so by using it in an IF statement you can achieve what you want.

[PerchloricAcid]

The part which is bugging me is

but I don't know how exactly would I "pick out" the values I want

how exactly would I read from the file.

For I moment, I had thought that the modulus might come in handy, if I could somehow get the program to know that each of the integer has a value related to it's order. This value would be one for the first integer, two for the second, etc. If I could do that, I would definitely use the modulus, but I figured last night that there might be something simpler, because I'm not really sure about how to designate those values.

[Steax]

Can you tell us more about what you've done before, or how far you've learned? It helps a lot.

Typically in your situation I'd do this, in psudeocode:

Code: Select all

loop = true;
i = 0;
while(loop){
   if(get_file_line(i) == EOF){ loop = false; }
   if(loop % N == 0){
      save_to_other_file(get_file_line(i));
   }

   line++;
}


(I'm leaving it in psudeocode so it doesn't get to instructive; it may not be the best way to do it. It's just how I'd approach it if I didn't know anything else about what it would technically require. You could also count the lines and do a for loop with a different increment, yes, if you'd like, but I prefer flexibility, and hence looping through every line. Also, I dislike break.)

[PerchloricAcid]

Thank you, I'll try to make use of what you wrote.

Well...
I know about loops, I know about classes (irrelevant here, I believe), I know... well, basically, I know stuff I needed for this simulation of mine, and a tiny bit more. So I'll just say what my program does.
I have a simplified three-body simulation which asks the user for initial conditions, saves them to a file, uses them for calculating stuff (using the RK4 integration method, but I've also toyed with other integration methods), saves the results in another file (t, x, y, vx, vy, ax, ay - for each moment defined by the time step dt).
Other things I know are just things I have come across while searching for my_simulation-related stuff, can't think of anything specific.

[Yakk]

A good trick is to make your format a bit more readable.

As an example:

Code: Select all

[INITIAL]
T:=
X:=
Y:=
VX:=
XY:=
AX:=
AY:=
[TIMESTEP]
T:=
X:=
Y:=
VX:=
XY:=
AX:=
AY:=


Your reading code now looks like this:

Code: Select all

1: Look for next line that is [INITIAL] or [TIMESTEP].  Lines that aren't [INITIAL] or [TIMESTEP] are ignored.
2: Look for a line that looks like VALUE:=VALUE -- ie, look for the :=.  Lines that don't contain := are ignored.
3: Evaluate what the part after the := is.  Include error handling if it makes no sense.
4: What is the part before the := -- is it "Y", "T", "X", "VX", "VY", "AX" or "AY"?  For each of these cases, set some value in your program.


Sure, this involves some really simple string manipulation. But your data files are now far, far more human readable. And if something goes wrong with your code (an extra line is included, or missing) it will be far more obvious what happened. Oh, and you can even extend your file format -- adding ENERGY:= to the [INITIAL], for example, or what have you.

The important part is that the file format is human readable, however. That'll help you fix more bugs than anything else.

[Rysto]

Ugh, I just found the following bug at work:

Code: Select all

    /* Note that we have to be very careful to not allow foo to reach 0.  If it does all hell will break loose! */
    while (foo > 0 && other_conditions()) {
         foo--;
         /* ... */

Yes, the comment was completely accurate.

[EvanED]

Code: Select all

"format ‘%s’ expects type ‘char*’, but argument 3 has type ‘int’"

is a fun warning. Especially when you look at it and find out that there's no way it could work.

At least it was only in "this will never run"-type code. (Which is probably why it lived.)

[PerchloricAcid]

Thanks for the help, guys.
I solved my problem
[However, later on I figured out that there was a more elegant way to do what I wanted, so I basically didn't even need this what I asked for, but practice is always good anyway.]

[Shivahn]

Hey, I'm having a ton of trouble with a very simple Scheme function. I want to make a function that takes a function and a number, and returns the function that performs that function on itself that number of times. In other words, given a function f and a number n=2, it should return f(f(x)). Given g and 5 it should return g(g(g(g(g(x))))). Here is my code:

Code: Select all

(define (repeat f n) (lambda (x) (if (= n 1) (f x) (f (repeat f (- n 1))))))
(define (square x) (* x x))

The square stuff is there for the testing. Now, when I type in ((repeat square 2) 5) what I expect to happen is for the interpreter to perform square on the value of square 5, which should be 25, so it should return 625 total. Instead what I get is "*: expects type <number> as 1st argument, given: #<procedure>; other arguments were: #<procedure>". That's within the body of the square function. When I go through the debugger, I notice that the repeat function is not recursively calling itself, so presumably it's returning the function (square (repeat f (- n 1)) or something similar. Am I right in thinking this? Does anyone know how to force the repeat function to recurse? And more importantly, can someone explain to me why it's not? (If that's what it's doing, of course.)

[Yakk]

First, learn how to indent. Code on one line is not easier to read, and code should be written to be read.

Code: Select all

(define
  (repeat f n)
  (lambda
    (x)
    (if
      (= n 1)
      (f x)
      (f
        ( repeat f (- n 1) )
      )
    )
  )
)


So, what does this do? It returns a lambda that takes 1 parameter. This lamba then evaluates if n equals 1. If n equals one, it returns f(x).

If lambda does not equal 1, it returns (f (repeat f (- n 1))). First, notice the lack of an x in that expression.

Second, what is the return type of (repeat f integer)? You are explicitly calling f with this argument. Write out the type. Is there any way you can get from this type to the type you want?

(Do you know how to write out types of functions, like in mathematics?)

[Shivahn]

Sorry about the indentation. I have a really terrible habit of ignoring good practices when writing up smaller bits of code.

I fixed it by making it

Spoiler:

Code: Select all

(define
  (repeat f n)
  (define
    (repeath f n x)
    (if
     (= n 1)
     (f x)
     (repeath f (- n 1) x)
     )
    )
  (lambda
    (x)
    (if
      (= n 1)
      (f x)
      (f
        ( repeath f (- n 1) x)
      )
    )
  )
)

But I'm not sure I understand everything you said, or did this the "right" way, so I want to be clear. Before, I know I was calling f on the function returned by (repeat f integer). I feel like I'm missing something with the type - isn't the type returned simply a function that (takes a function that takes a function) that takes x? I don't know what I changed here, other than making it so that it's using the same x all the way down. I feel like I'm missing something critical now.

And I don't know how to write out types of functions. While I've coded off and on for some time, it's only now that I'm getting into the theory and actual meanings behind the words and academics and all, so my background knowledge can be pretty lacking in certain areas. If there's a quick tutorial you have or something, I'd love to read it.

[Yakk]

So, when you have a function from Real to Integers, the function's type is R->Z (where R is real, and Z is integers). You'd write this as f:R->Z in mathematical notation.

In C++, the function might be declaired as R foo(Z), or auto foo(Z)->R -- that last one being more similar to mathematical notation.

What you did works, right? You now feed x back into the lambda you get from your recursive call.

The type of your repeat is then:
(A->A, N)->(A->A)
where A is some unknown type, and N is your "repeat count".

A more amusing way to write repeat could be:
(A->A)->(N->(A->A))
which is logically similar, except now repeat takes a function from A to A, and returns a function that takes the number of repeats and outputs that function from A->A.

In a sense, this takes f(x) as input, and outputs the function g, where g(n) = f^n.

Another amusing way to write repeat is like this:
N->((A->A)->(A->A))
which would be a repeat function that takes a parameter of the number of repeats, and returns a function that takes an A->A function and returns a function that repeats it n times.

This repeater generates the "to the power of n" operator.

Writing both of these could be illuminating.

[EvanED]

The type of your repeat is then:
(A->A, N)->(A->A)
where A is some unknown type, and N is your "repeat count".

Better would be to parenthesize it as

Code: Select all

( (A->A), N ) -> (A->A)

Pairing binds tighter than function application in every typed functional language I know, so (A -> A, N) would be interpreted as a function of one argument that returns a pair. While there's nothing fundamentally wrong with doing the other thing, it goes against expectations heavily. At least for people who have the background to have expectations.

A more amusing way to write repeat could be:
(A->A)->(N->(A->A))
which is logically similar, except now repeat takes a function from A to A, and returns a function that takes the number of repeats and outputs that function from A->A.

This is called Currying, after Haskell Curry. (The same dude that the Haskell language is named off of.)

Unfortunately, Lisps aren't a very good language for curring in without writing or using some supporting code to convert a non-curried function to a curried one. I'm not saying don't write them, but expect it to be annoying. For instance, you'd have to say ((repeat square) 3) instead of (repeat square 3).

[headprogrammingczar]

Not strictly relevant, as the problem has already been solved, but lispy indentation style goes more like this, with the lisp programmer reading indentation instead of parens:

Code: Select all

(define (repeat f n)
  (lambda (x)
    (if (= n 1)
      (f x)
      (f (repeat f (- n 1))))))


[EvanED]

FWIW when I was doing a bunch of Lisp programming I came to despise the "lispy" indentation style and usually use something more like Yakk's. The "indentation is all you need" is a bit of a pipe dream, and it makes editing harder.

But this is more a conversation for the religious wars forum.