/dpt/ - Daily Programming Thread
- Home
- Board: /g/ - Technology
- Reading: /dpt/ - Daily Programming Thread
Old /dpt/: >>51488679
What are you working on, /g/?
Rate my strrev implementation.char *strrev(char *str)
{
int length = strlen(str) - 1; /* omit \0 */
char *tmp = malloc(length);
int a = length;
int b = 0;
while (a >= 0 && b <= length)
{
strncpy(tmp+b, str+a, sizeof(char));
a--;
b++;
}
return tmp;
}
ITT: millenials get BTFO by basic CS question.
LMAO
LOL
KEK
>>51491633
What is the basic CS question?
>inb4 "do you wear a skirt while programming?"
Are you ready for round 2, /g/?
>protip: you aren't
>>51491650
averaging two integers on C
>>51491633
>>51491681
samefag
please get out and stop trolling this thread
>>51491681
what is that, likeint avg(int a, int b)
{
return (a * b) / 2;
}
??????
>>51491691
>muh animu
>muh traps
>muh degeneracy
>muh jerking off to kids
pedo...
Studying java for a year, and this problem beats me
Given a M * N matrix, print the minimum value for each row and each column, but this has to be a separate method, to which you pass a matrix of integers
eg: minValues(m) and the output would be
-min for row 1: 5
-min for column 1: 3
-min for row 2: .... Etc
How do i do this?
>>51491716
yes
>>51491564
FAAAAAAAAAAAAAAAAAAAGGGGGGGGGGGGGGGGGGOTS
>>51491564
>What are you working on, /g/?
Averaging two ints in C
>>51491716
>*
No, average would be (a + b) / 2
Also, if a or b or both are high values, (a + b) might overflow causing undefined behaviour in C.
>>51491737
Please be joking.
>>51491747
It sounds easy but it's not. Go ahead and try
>>51491744
oh ok
just use size_t and return a size_t.
Why is this an issue.
You'd be more hard pressed to find people who can make a function that can take a variable argument list and create a function that isn't limited to 2 ints.
>>51491767
>Go
TRIGGERED
>>51491771
>Why is this an issue.
Because negative numbers exist.
>>51491780
>animooooooooooooo
degenerate, please return to your /a/
>>51491771
>just use size_t and return a size_t
size_t can't represent negative integers
>>51491747
>>51491737
That is not as easy as it sounds.
So, /g/, what should someone too smart for programming do?
ITT: anime faggots and trap lovers get triggered over basic CS discussion so they want to derail the thread again to anime and traps.
LOL
>>51491747
>>51491737
Tried that but if i pass a matrix that is not square, it throws an exception
need to write a java app 1 week to impress some people for a job interview.
What should i write?
using Java and Hibernate
>>51491863
Probably go back to fapping to ponies
>>51491863
Get into finance.
>>51491771
>just use size_t and return a size_t.
size_t can't represent negative integers
>Why is this an issue.
It's really not, just your average /g/ autism
>You'd be more hard pressed to find people who can make a function that can take a variable argument list and create a function that isn't limited to 2 ints.
I agree, but if you are asked that question in an interview you should say "oh, you can do (a + b) / 2 but you should be aware that this might overflow" or something similar.
>>51491911
>there's a correct way and a simple and shitty way
>I only know the simple and shitty
>I can't write reliable code
really, /g/?
>>51491911
>It's really not
>oh, you can do (a + b) / 2
looks like it's an issue for you
>>51491946
The correct way would be to cast to double, since the question in no way implies that you need to store/truncate the result into an int.
But I guess this too tickles your autism, doesn't it? And you're going to reply with a bunch of new shitposts too, aren't you?
>>51491771
>just use size_t and return a size_t
orly? and how would that work with 1 and SIZE_MAX? is the average 0?
fucking moron
>>51491964
>only quotes half of the sentence
Also, see >>51491982
>inb4 autistic shitstorm
>>51491982
>I don't know how to do it with ints
>I need muh dublez!
fucking webdev amateur
>>51492000
>using webdev as an insult
ok kid
You do know that floor(avg) is not actually the average, right?
>>51491965
is this your homework? how can you be this lazy?
>>51492016
Its for work. And i am not lazy, i just cant do it
>>51491600
>>51491591
destructive versionvoid strrev(char *str) {
size_t length = strlen(str);
char *a = str,
*b = str + length,
tmp;
while (a < --b) {
tmp = *a;
*(a++) = *b;
*b = tmp;
}
}
>>51492000
>doesn't care about precision
Found the NEET
>>51491982
>in no way implies that you need to store/truncate the result into an int.
>what is gmp
>what is mpz_add
>what is mpz_tdiv
anon, can you even bigint?
>>51492014
>floor(avg)
>average
webcuks actually believe this!
>>51492032
>>51491737
>print the minimum value
>has to be a separate method to which you pass a matrix of integers
your job would not require you to do something like have a separate method. this is for your class.
>>51492037
>cares about precision of integers when he needs to determine an integer between two other integers
how's that css coming along, boy?
>>51492040
>GNU extensions
Do you even know C?
>>51492081
>extensions
this ain't CSS, webfag!
>>51492073
>an integer between two integers
That's not the average, anon
That can be calculated in this way:int between = MIN(a, b) + MAX(a, b) - MIN(a, b); // yes this overflows too, I'm doing this on purpose
>>51492111
GNU aint C kid, good luck trying to get that shit into the Linux kernel
>>51492112
You're a moron. That's just MAX(a, b).
What the fuck is going on here
>>51492129
Good luck trying to get CSS and JavaScript in the Linux kernel, webfag.
>>51492136
Found the web developer
It's actually potential demons flying out your nose.
>>51492141
Just webfags getting rekt by a big-dick-C-god.
>>51491600
if you're not allocating space for the null terminator, aren't you returning a pointer to an unterminated string?
>>51492141
millenials and webdevs getting BTFO by C kun
>>51492081
>GNU extensions
GMP is a library
What is the stack in C, anons?
>>51492060
>this is for your class.
Well yes, he did say it was in Java after all.
>>51492169
>>51492182
samefag
Has anyone ever purchased international versions of textbooks? Are there any real differences?
The US version of a book I'm looking at is $160. The international version is less than $20.
>>51492212
that's even worse
Hi, I don't want to open a new thread for a stupid question.
So, I'm already a programmer (C++,Java,PHP) and I want to learn Python.
What is the fastest way?
>>51492229
you cannot buy the international versions of those books because they weren't made for FREEDOM
http://yro.slashdot.org/story/12/04/16/1843257/student-charged-for-re-selling-textbooks
>>51491879
need some help fuckers.
thx in advance
>>51492229
If you don't intend to keep the book, as is the case with most college texts, it's fine.
They're usually black and white and have shittier quality paper stock, but the content is the same.
Just make sure they're brand new, the price difference is often negligible.
If you're looking to buy international editions of meme programming books you intend to keep on your shelf, please don't.
I bought an indian copy of K&R once and the paper smell was so unbearable I had to throw it away.
>>51492239
>I didn't know, I'm just a webdev
is Perl good?
>>51492350
>I use GNU libraries because I'm too stupid to average two integers in pure C
I'm not the one who need a third-party library to average two numbers.
>>51491879
can't you research what you employer bussiness is and do a small app that would help him in his business domain?
>>51492390
>libraries
>extensions
>they are the same thing to me because I only know HTML
ishyggydiggy
>>51491564
LINK TO THE NEW THREAD FROM THE OLD ONE FUCKING RETARDED FAGGOT
>>51492314
>I bought an indian copy of K&R once and the paper smell was so unbearable I had to throw it away.
That's terrible. I tend to buy the textbooks for that are part of my major courses to keep as reference; so if they are shit quality, then I won't bother.
>>51492409
>literally extending your application with additional libraries
>not extensions
>needs libraries to average two integers
>thinks "web development" is HTML and CSS
Found the NEET
>>51492431
IT DIDN'T HAVE AN ANIME IMAGE. HOW WAS HE SUPPOSED TO SEE THAT? FUCK YOU NIGGER
I see you're still arguing over C, maybe some of you (doubt it tá Žbá Žh) can tell me what's wrong with this C code. Some anon said it's unreliable:void copy_stream(FILE *in, FILE *out)
{
int c;
while ((c = fgetc(in)) != EOF) {
fputc(c, out);
}
}
>>51492431
DON'T POST A NEW THREAD BEFORE THE BUMP LIMIT FUCKING RETARDED FAGGOT
THE BUMP LIMIT IS 310
>>51492450
>>thinks "web development" is HTML and CSS
he's right tho
>>51492477
>that pic
literally not an oop problem
two fucking integers or one being used as two
>>51492443
>to keep as reference
What's your major?
Also, if you want to avoid the smell thing, make sure they're brand new.
They come sealed in plastic.
That K&R was $4 shipped, and I didn't notice it said used.
>>51492390
>My language of choice coerces all numbers to doubles anyway
>>51492450
extension means a different thing in c
libraries are portable. compiler extensions are not (necessarily).
>>51492478
this desu
you can make a thread early if you've got a new epic OP pic
>>51492493
>"he"
lol
>thinking HTML and CSS is programming
lol
>>51492477
c is not initialized
>>51492521
>thinking web development is programming
anon...
>>51491875
> protip: read the exception, figure out why it gets thrown and what you can do about it, then either adapt or completely change your code. In brief be a big boy stop crying, and do real programming
Or if you are a trans woman of color, we'll help you, we promise, but POST YOUR FUCKING METHOD, A MATRIX THAT CAUSES THE EXCEPTION, AND THE DETAILS OF THE EXCEPTION BLOODY HELL!!!!
>>51492477
honestly idk but what would happen if in and out are the same file? infinite writing?
>>51492521
kill yourself
>>51492506
>My language of choice coerces all numbers to doubles anyway
It doesn't, which is why you need to cast it to two doubles in order to get the average.
>extension means a different thing in c
Wrong. I didn't say "compiler extensions", I simply said extensions.
The library web page even use that exact same word about itself (and about the more complete library, MPFR).
>libraries are portable
Not necessarily
>>51492555
>infinite writing?
no
>>51491737
>Studying java for a year
for like 1 hour once a month?
>>51492541
>>51492565
>hurr durr HTML is programming
>herp derp oh wait i didn't say it was durrr
>codeeval asks you to efficiently get all the prime numbers up to a certain integer
>your code wastes time calculating the largest one first so you know where to stop putting commas in the output
>all because you're too lazy to dynamically allocate a string buffer and replace spaces with commas.
Should I even bother?
>>51492569
webcuk detected
>>51492582
>animu
m8...
>>51492585
Always bother unless it's not worth bothering
>>51492496
friendly reminder that you handled a book that had been used by people who don't poo in the loo and who wipe themselves with their bare hands
>>51492587
>needs a third-party library to get the average of two numbers
>doesn't give a fuck about portability
>doesn't give a fuck about performance
>doesn't give a fuck about ease of maintenance, ease of deployment etc
>doesn't give a fuck about memory or disk usage
I'm fairly sure you're the "webkek" here, who else in their right mind would include a library to do basic arithmetic?
>>51492477
Too soon, anon, too soon!
Stream IO is a far too advanced topic for /dpt/, these retards can't even average 2 numbers nor do they know how to use memcpy. Maybe in 3-4 years, at the rate they learn.
>>51492626
don't fucking remind me
>>51492525
>c = fgetc(in)
>>51492585
Do the first one, then do every other one putting a comma before them
>>51492628
>muh grunt and gulp are standard tools
>muh nodejs is webscale
>muh madness
>I'm a htmlcá Žuck
>>51492477
Jesus Christ, no one called it yet?
You don't write the terminating EOF to the stream you're copying to.
>>51492676
>no one called it yet?
this is the level of /g/
>>51492654
>knowing about webtools
>"I'm a htmlcá Žuck"
Glad you admit it.
>>51492683
>proud about not knowing
anon, I hate to break it to you, but you're an ignorant moron
>>51492697
At least I know how to average two ints without having to rely on large third-party libraries
>>51492697
>proud of being a web code monkey and showing off his "knowledge" about the latest hipster frameworks
>>51492676
You don't write EOF to a file. You fclose it and that's that.
>>51492680
Shut the fuck up.
>>51492758
Okay, he doesn't close/flush the stream then.
>>51492797
Why would a function, given an open stream as input, be expected to close it?
>>51492797
that's the responsibility of the caller of the function really
Can somebody help me understand why C's bsearch() works this way? Pic shows the function definition#include <stdio.h>
#include <stdlib.h>
#include <string.h>
int dumbcmp(char **, char **);
#define NUM sizeof(wlist)/sizeof(char *)
#define WIDTH sizeof(char *)
int main()
{
char *wlist[] = {
"aids", "crud", "evil", "gall", "idiot", "kick",
"moon", "oval", "quod", "sulk", "under", "west",
"year"
};
char *word = "evil";
int found;
/* why doesn't this work... */
found = (bsearch(word, *wlist, NUM, WIDTH, strcmp) != NULL);
printf("%d\n", found);
/* ...but this does? */
found = (bsearch(&word, wlist, NUM, WIDTH, dumbcmp) != NULL);
printf("%d\n", found);
}
int dumbcmp(char **a, char **b)
{
return strcmp(*a, *b);
}
>>51492676
It won't, it exits the loop as soon as it reads EOF
>>51492797
But what if copy_stream() isn't meant to close a file after writing to it? What if it's to be designed to allow one fopen(), multiple copy_stream()s and one and one fclose()?
I don't think there's anything wrong with his function, but that's just me.
>>51492817
>>51492814
It would be expected to flush it.
>>51492840
Why?
>>51492847
Because of how I/O streams work
>>51492858
Flushes are expensive. Why do one if it's not strictly necessary?
>>51492871
>not strictly necessary
Consistency is strictly necessary
But the real question is if you think flushes are expensive, why would you read and write single bytes rather than reading a whole chunk into memory and writing that chunk to the stream?
>>51492840
Not necessarily. printf() does not always flush (it only completely flushes if the last character you pass to it is a '\n', or if the number of bytes it has to write to stdout exceeds the size of its internal buffer).
The problem in his implementation lies on whether or not in == out, I believe. If that is the case, then reading half of the file + writing it to itself will cause the cursor to reach the end of the file, meaning that the last half of your file gets destroyed. Depending on whether the number of bytes in your steam is even or odd, you might also go on trying to write in junk memory.
>>51492904
Beats me.
who /java/ here?
>>51492917
>The problem in his implementation lies on whether or not in == out, I believe.
Well, reading and writing to the same file simultaneously is no problem as long as you do open it with the rw modifier (and not r+)
>>51492958
I was working under the assumption that the file was opened for rw. I believe that rw keeps one single cursor though, so a read increases the same cursor a write this, and not separate ones (e.g. fgets followed by fputs places your character at offset 1 instead of offset 0). I don't know if this is implementation-defined though, but that's how I think it works for 64 bit linux.
>>51492951
Most people actually enjoy programming
>>51492951
this.
who the fucks cares about autistic C garbage like averaging integers.
most employeers don't need library programmers.
>>51493034
What's wrong with Java?
>>51492904
>But the real question is if you think flushes are expensive, why would you read and write single bytes rather than reading a whole chunk into memory and writing that chunk to the stream?
I've limited knowledge of C but doesn't fgetc work with streams underneath? Which means while you're reading it char-by-char it's actually buffered internally.
>>51493000
>>51493000
Actually, rw is non-standard.
a (a+) allows reading from start and writing from end.
>>51492822
>string literal to char*
you are one step away from breaking the universe
there is nothing wrong with java it's literally the best language along with C++
>>51493045
it is shit
>>51493057
Yes, from an I/O perspective you're quite right, but doing two function calls per bytes is in-efficient if you're copying a lot of bytes, and you're also totally at the mercy of the stream implementation (which is free to do a syscall for every call to fputc for example).
>>51493058
Well then, aside from the case where in == out and the file pointer has permission to both write to and read from itself, I believe anon's function holds up.
Oh, there's also the case where he leaves old characters on a file if he copies an amount of bytes lesser than the target's size, but I believe that's expected to be taken care of by the caller.
>>51492711
>know how to average
last two threads prove otherwise, webshit
>>51492222
underrated
>>51492822
pretty obvious actually
when should i pass by reference vs pass by value in C?
if i'm not going to modify the variable, why pass it by reference?
>>51493096
>last two threads prove otherwise, webshit
I didn't post in the last two threads.
>failing to tell completely different people apart
lol
>>51493045
no unsigned integers
Shitty generics
No easy way to box primitives. (have to deal with bullshit like int and Integer types)
no pointers
painful native code interop
the most verbose language in popular use by far
Currently developed by Oracle.
probably loads of other shit i can;t remember
>>51493118
>if i'm not going to modify the variable, why pass it by reference?
Because it is "cheaper" for compound types.
Lets say you have a struct with many members. It's easier to pass it as a const struct something* than as a value-type.
>>51493118
Passing a large object by value can incur an expensive copy. A good rule of thumb is to pass by const reference if you're not going to modify it.
>>51493118
There's no "pass by reference" in C. You always pass by value. What might be the case is that your value is a pointer.
All arguments must be put on the stack. Passing a pointer means you must put an address on the stack. Passing whatever it points to means you have to put that entire thing on the struct. This matters for huge unions and structs, mostly.
>>51493124
>pretending
back to reddit, csstard
so what is actually the correct way to get an average of two ints in C?
>>51493124
> I am not anon, I am anon, DUH!
>>51493131
It really is verbose. I've started learning it few months ago, and even the simplest shit that I want to make takes lines and lines of code.
>>51493147
>All arguments must be put on the stack
No, they can be put in registers.
>>51493131
boohoo
>>51493149
>still going at it
you lost, go include some bloated library for adding two integers
>>51493131
pretty good list. I would add forced pass by value and inability allocate objects without references (Object o; in C++)
>>51493151
apparently nobody on this shit vietnamese blog knows
>>51493142
Explain that...
A const reference? Does that mean I can't change the variable?
Can you give me an example?
>>51493151
http://ideone.com/qXPpZQint avg(const int a, const int b) {
long long difference = (long long) b - a;
if(b < a) {
difference += difference & 1;
}
return a + difference / 2;
}
>>51493159
>assuming anyone would spend three /dpt/s arguing with a sperglord
>>51493176
>pls no bully
>I can't take any more ass raping from your big dick
>>51493168
also this, retarded user base
>>51493131
>no unsigned integers
Not really a problem.
>Shitty generics
This is true.
>No easy way to box primitives. (have to deal with bullshit like int and Integer types)
Well, it's pretty easy to make your own holder type, but this is a fair enough complaint.
>no pointers
Not a problem.
>painful native code interop
This is a true complaint, but not really valid (it's not what java was meant to do).
>the most verbose language in popular use by far
Annoying, but not really a huge problem (I don't mind it, especially considering that an IDE makes it relatively effortless to write java)
>Currently developed by Oracle.
Irrelevant
>>51493195
orint avg(const int a, const int b) {
long long difference = (long long) b - a;
difference += difference & (b < a);
return a + difference / 2;
}
>>51492646
This works, thanks.
Also, I noticed it runs much faster if you stop checking for divisibility at sqrt of the number of you're checking.
Why is that?
>>51493195
implementation defined as fuck
>>51493147
>All arguments must be put on the stack
that is implementation defined. It can be easily optimized by a compiler
Also C has pointers, so pass by reference is not really a necessity.
>>51493193
Since you're talking about C you have to use a const pointer.typedef struct Big {
int stuff[1024];
} Big;
// makes a copy even if it's not strictly necessary since you're not modifying it
passByValue(Big big);
// doesn't make a copy, retains semantics of immutability
passByConstPointer(const Big *big);
>>51493193
>Does that mean I can't change the variable?
Yes
>Can you give me an example?void print_string(const char* str)
{
// i can read str here, but I'm not allowed to write to it
while (*str != '\0') {
char c = *str; // this is allowed
*str = c; // this is not allowed [it is undefined behaviour]
str++;
}
}
>>51493225
incorrect
Is it possible to go more enterprise than this: https://en.wikipedia.org/wiki/Dependency_inversion_principle
>>51493166
You're right, but if the thing you're referencing is big enough for this to matter, it's probably not going to completely fit on the registers.
>>51493232
>>51493251
no FOR FUCK'S SAKE RETARD
>Long long signed integer type. Capable of containing at least the [−9223372036854775807, +9223372036854775807] range;[3] thus, it is at least 64 bits in size. Specified since the C99 version of the standard.
>>51493211
You can take some more anon, my dick isn't that big.
>>51493195
what if int and long long have the same representation
i'll admit i don't think such a thing can actually happen right now but the standard doesn't disallow it
>>51493252
It's just an over engineered as fuck example of "program to an interface, not an implementation".
>>51493252
starship tier enterprise
>>51493252
>Is it possible to go more enterprise than this:
Yes
Including a bigint library in order to calculate the average of two integers, for example.
>>51493251
http://ideone.com/HhYyR9
these two are perfectly equivalent FUCKING RETARD
>>51493195
>>51493225
https://github.com/woky/chrome-mouse-wheel-tab-scroll
>>51493061
The original function where this problem arose used a passed char* as the key instead of a string literal. I'm just simulating it here.
>>51493107
pls explain. It looks like bsearch() is passing strcmp() a pointer to each array element (char**), instead of the element itself (char*) which strcmp expects. Is that the expected behavior? If so, why?
>>51493306
>equivalent
Yes, correct, no?
How is truncating the average into an integer value "the average of two integers"?
>>51493227
It should be obvious why it's faster, you're only doing "sqrt n" iterations instead of "n".
It works because as soon as you go past the square root, any factor you find would be the counterpart of a factor that you've already checked. If you haven't found any factors yet, there cannot be any more.
>>51493283
then you use stdint with int32 and int64. or an arbitrary size integer representation for a-b. it's just a toy example and you know it works in practice because no one expects ints to have more than 32 bits
>>51493324
autism fuck you i hope you die in a fire
the average of two integers is a truncated integer unless otherwise specified FUCKING AUTIST I SERIOUSLY HOPE YOU DIE
>>51493330
What if you use C89/90 where there's no such thing as stdint.h?
>>51493346
A U T I S M
U
T
I
S
M
>or an arbitrary size integer representation for a-b
>>51493330
>then you use stdint with int32 and int64.
but then you're deviating from the spec of the average of 2 ints :^)
>>51493324
oh, didn't notice that you have autism. my bad
>>51493343
>the average of two integers is a truncated integer unless otherwise specified
Wrong
https://en.wikipedia.org/wiki/Average
>In colloquial language, an average is the sum of a list of numbers divided by the number of numbers in the list. In mathematics and statistics, this would be called the arithmetic mean.
https://en.wikipedia.org/wiki/Arithmetic_mean
Do you define PI as 3 as well, because it wasn't specified as something else than an integer?
>>51493365
how would you write an average function anon?
>>51493364
>>51493357
>>51493343
>webcuk defines PI as 3 because precision wasn't specified
>>51493365
why are you in a programming thread
>>51493299
>producing code with rounding errors
repugnant desu senpai
>>51493384
if you knew anything about this subject you would know that all numbers that most webcuks use are double-precision floats
>>51493382double average(int a, int b)
{
return ((double) a + (double) b) / 2.0;
}
>>51493384
>autistic defines PI as a rational number
>>51493084
>doing two function calls per bytes is in-efficient if you're copying a lot of bytes
Function call overhead is negligible compared to the cost of IO
>you're also totally at the mercy of the stream implementation (which is free to do a syscall for every call to fputc for example)
Same thing applies to fread/fwrite. fread/fwrite are usually implemented in terms of of fputc/fgetc rather than the other way around.
>2015
>not storing all your data in Pi
have fun worrying about disk space
>>51493396
>truncating a fraction into an integer =not rounding error
>using floating point precision = rounding error
>>51493385
>programming = incorrect math
wut
>>51493406
>no u
lame comeback, anon
>>51491737
lol this is so your homework. and this is easy homework kid
>>51493438
Let's see how you would do it anon
>>51493414
what if double can't hold all integers?
>>51493420
>Function call overhead is negligible compared to the cost of IO
Yes, but that doesn't mean that it's cost free. If you're on the track of doing performant code, you should think about this.
>Same thing applies to fread/fwrite. fread/fwrite are usually implemented in terms of of fputc/fgetc rather than the other way around.
Yes, but there's a vast difference between doing a syscall per X bytes and doing a syscall per byte.
>>51493438
>average two ints
>no further specification
>autistically assume for no valid reason that a function that takes ints as input should return a float
>>51493438
not an error if you do it on purpose
>>51493414
AUTIST fucking kill yourself i'm dead serious
>>51493454
See >>51493414
>>51493458
Then you implement your own number representation using strings.
>>51493489
kill yourself
>>51493267
int can be 64 bits too, it's also specified in the standard
>>51493469
>calculate average
>average
>autistic to assume that a function that calculates the FUCKING AVERAGE would return a floating point number
lolwat
>>51493486
I'm sorry you failed the job interview.
>>51493489
So your answer gives wrong answers? lol what a retard.
>>51493514
then >>51493414 is also wrong. CHECK MATE AUTISTS
>>51493531
doubles can be 128 bit
>>51493531
It is.
>>51493518
kill yourself
it's a function that averages two ints. my function returns the average of the two ints with full integer accuracy. there is no reason to assume that it should return a double. YOU FAIL
kill yourself
oh yeah well I got some doubles for you right here
>>51493546
int can be 256 bit
>>51493555
>my function returns the average of the two ints with full integer accuracy
That's not the average though.
>there is no reason to assume that it should return a double
Yes, it's the fucking average, not the average rounded down to the nearest integer.
Do you think the square root of an integer should also be returned as an integer?
>>51493567
your mom can be googolplex bit
>>51493166
>>51493147
> What might be the case is that your value is a pointer.
You mean by value is a memory address basically?
> All arguments must be put on the stack.
Stack?
Question:
What's a better practice, keeping a global extern variable to store data, or passing around a reference in the arguments?
guys calm down, i got this
bool average(int a, int b)
{
return ((a & 1) != (b & 1));
}
>>51493414
I'm pretty sure a bunch of people did this last thread and you told them they were wrong
>>51493577
>Yes, it's the fucking average, not the average rounded down to the nearest integer.
your answer is also rounded.
>Do you think the square root of an integer should also be returned as an integer?
yes.
>>51493584
yes; bottom line, the code is wrong and nobody presented a good solution for averaging 2 ints
>>51493595
>keeping a global extern variable to store data
Never ever use a global mutable variable.
>>51493555
>calculating the average as a double is autism
>you should rather include a third-party library to handle bigints
ITT: webdevelopers programming in C for the first time
>>51493599
it returns true if there will be a rounding error
false if there won't
problem solved
>>51493607
STUPID ASS TROLL GO FUCK YOURSELF
>>51493601
>you
I'm not that guy, see the replies to >>51491982
>>51493602
>your answer is also rounded.
True, but it's not truncated entirely. But yes, there will be precision errors for some values.
>>Do you think the square root of an integer should also be returned as an integer?
>yes.
Okay then.
>>51493607
LOL
I see everyone is busy showing their broken average implementation and ignore telling >>51492477 why the code is not correct.
That code is not guaranteed to copy the entire input stream because it stops at EOF instead of stopping when end-of-file or an error is encountered. Use feof() and ferror() to detect those cases. The function should also return an error code to signal those errors.
>>51493607
NO IT'S THE EXACT OPPOSITE OF WHAT YOU'RE SAYING
WEB DEVS AND STUPID ASS KIDS LIKE YOU WITH NO PROGRAMMING EXPERIENCE WOULD ASSUME THAT IT'S FINE TO RETURN A DOUBLE WHEN THE SPEC CLEARLY IMPLES AN INT.
>>51491681
For unsigned: small + (big - small) / 2
Parity of (big - small) tells you if the average can be represented accurately as an integer. Adding half the difference rather than dividing the sum by two avoids overflow.
For signed: if both are on same side of 0, use unsigned procedure, else use (a+b)/2 and parity of (a+b).
You can probably avoid the branches if you rely on the twos complement representation.
Easier, fine for most purposes: convert to double, (a+b)/2
If you aren't using your largest integer size: convert to i64, (a+b)/2.
Seems like there should be a better way to do signed, still without relying on larger types, floats, or twos complement representation.
>>51493644
but truncated errors are also precision errors.
>>51493467
It fread/fwrite are calling fgetc/fputc in a loop anyways (they probably are) there is not going to be a measurable difference.
>If you're on the track of doing performant code, you should think about this.
Right, and you should know that it doesn't matter when your code is IO bound. And in general, you should profile before you even think about complicating your code to improve performance.
>there's a vast difference between doing a syscall per X bytes and doing a syscall per byte.
At the end of the day, you just have to hope that your libc devs aren't retarded. If you're that concerned about it, you should be using read/write directly.
>>51493666
>WEB DEVS AND STUPID ASS KIDS LIKE YOU WITH NO PROGRAMMING EXPERIENCE WOULD ASSUME THAT IT'S FINE TO RETURN A DOUBLE WHEN THE SPEC CLEARLY IMPLES AN INT.
>clearly
>implied
lolwat
>calculate the average of two integers
Clearly that implies calculating the arithmetic mean of two integer values, which is not floor(mean(x, y))
Why would you assume that a function must have the same return type as its parameters?
Do you think strlen returns a char pointer? Do you think sqrt returns an int?
what's wrong withint avg(int a, int b) {
return (a/2 + b/2) + (a & 1 || b & 1);
}
>>51491982
this is what normies actually believe
>>51493667
>if you rely on the twos complement
>fine for most purposes: convert to double
>convert to i64
implementation-defined/10
>>51493041
>averaging integers.
>library programmers.
Is this what programming has come to?
>>51493696
>which is not floor(mean(x, y))
right, it's trunc(mean(x, y))
>>51493696
REEEEEEEEEEEEEEEEEEEEEEEEEEEEETAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAARD
if the spec was to calculate the average of two doubles you'd return a double unless otherwise specified
if the spec was to calculate the average of two floats you'd return a float unless otherwise specified
if the spec was to calculate the average of two intsyou'd return a int unless otherwise specified
int average(int a, int b)
{
return (a + b);
// shr to get truncated
// &1 to determine if .5
}
nobody will notice one missing bit
>>51493718
The first two procedures aren't implementation defined, they use branches to avoid relying on twos complement.
I just mentioned some other ways for completeness.
>>51493707
it doesn't always work
>>51493691
>It fread/fwrite are calling fgetc/fputc in a loop anyways (they probably are) there is not going to be a measurable difference.
But they're not (unless you have an retarded implementation of fread/fwrite).
>Right, and you should know that it doesn't matter when your code is IO bound.
Not doing a syscall every byte anymore = no longer IO bound
Then it starts to matter.
>And in general, you should profile before you even think about complicating your code to improve performance.
Doing an fread into a buffer is not "complicated code" anon.
>At the end of the day, you just have to hope that your libc devs aren't retarded.
Yes, of course, but even if they're not, calling a function (or in this case, two functions) for every single byte should raise some eyebrows.
>If you're that concerned about it, you should be using read/write directly.
Not necessarily, calling a function for every byte should be a huge red flag for any developer.
>>51493675
One is more accurate than the other though.
>>51493763
one is more true to the spec though
>>51493746
that's a sum, anon
So /dpt/, did you learn a functional language yet or are you still only have limited skills as a developer?
>>51491681
Here's some solutionsint avg2(int x, int y) {
int sum = (x / 2) + (y / 2);
int mod = (x % 2) + (y % 2);
sum -= (sum > 0) * (mod == -1);
sum += (sum < 0) * (mod == 1);
return sum + mod/2;
}
int avg(int x, int y) {
int sum = (x / 2) + (y / 2);
int mod = (x % 2) + (y % 2);
int nsig = (sum < 0) - (sum > 0);
return sum + nsig * (nsig == mod) + mod / 2;
}
int32_t avg3(int32_t x, int32_t y) {
return (((int64_t)x)+y)/2;
}
int avg4(int x, int y) {
return (x / 2) + (y / 2) + (((x / 2) + (y / 2) < 0) - ((x / 2) + (y / 2) > 0)) * (((x / 2) + (y / 2) < 0) - ((x / 2) + (y / 2) > 0) == (x % 2) + (y % 2)) + ((x % 2) + (y % 2)) / 2;
}
avg3 is by far the fastest but not supported on all architectures.
the other ones are more or less the equivalent.\
didn't write these myself, but you can use them if you need a solution for this.
>>51493778
>can't read
>>51493784
oh and they're branchless.
>>51493775
>the spec
The spec says
>calculate the average of two integers
One calculates the average (albeit with less than ideal precision), the other adds two integers and then does long division.
Which one is the most "true to the spec"
>>51493741
See above.
>>51493763
>One is more accurate than the other though.
implementation-defined, what if integers are a thousand bit while double mantissa is only 10 bit
only correct solution is returning a struct that contains truncated value of average in int and a bool to store if it is 0.5 or not
>>51493696
>Do you think sqrt returns an int?
sqrt returns the same type that's input into it
sqrt takes a double, returns a double
sqrtf takes a float, returns a float
sqrtl takes a long double, returns a long double
don't think c has at square root function that takes an int but if there was one i'd expect it to return an int
>>51493760
when
>>51493799
it returns a+b, you fucking moron
>>51493195
retard here
what does "long long" do?
what does "difference & 1" do?
>>51493414
>Writing a binary search function
>use the average(int, int) function from /g/'s C library for conciseness
>my shit's all fucked up because anon tricked me into using a double as an index
>>51493806
https://www.youtube.com/watch?v=eBPqksG9nbA
the original spec was
>average 2 ints in C
and that's it
and using the word "integer" instead of "int" doesn't at all justify returning a double. why not say "number" then if the spec wanted a double?
you're a FUCKING RETARD who just wants the easy way out with the (a + b) / 2 solution FUCKING PATHETIC FUCKING STUPID ASS SHITKID
>>51493825
maybe you should read the comments
they are there for a reason
>>51493822
>don't think c has at square root function that takes an int but if there was one i'd expect it to return an int
Autism
>>51493832
I'm pretty sure C won't let you index with a double, since it desugars to an offset (which would be a type error) and dereference.
>>51493822
>when
-1 and 0
>>51493855
>tricked you
back to StackOverbanana codemonkey
>>51493855
>binary search
>average
Why would you do that unless you're implementing an incorrect binary search?
>>51493858
>the original spec was
>>average 2 ints in C
>and that's it
Yes, and the average of two ints is a fraction.
>you're a FUCKING RETARD who just wants the easy way out with the (a + b) / 2 solution FUCKING PATHETIC FUCKING STUPID ASS SHITKID
You can still do the "easy way out" by simply casting to long long, you have no point at all.
>>51493832
a long long int is at least 64 bits wide (in C99 or later) so it can hold the sum of any two 63 bit or smaller integers. "difference & 1" masks the least significant bit so it's 1 if difference is odd and 0 if difference is even.
>>51493860
the comments don't make the function return the average; it still returns the sum
>>51493832
long long is a type like int
int is guaranteed to be at least 16 bits, long long is guaranteed to be at least 64. but nothing prevents them from being the same (on most modern implementations int is usually 32 bits and long long is 64 bits, but that's not guaranteed)
difference & 1 is a bitwise operation. i'm shit at explaining so just lookup "bitwise and"
>>51493855
For binary search of an array you can use small + (big - small) / 2, since you don't need a branch to find the bigger/smaller, you're always dealing with unsigned numbers, and the parity doesn't matter.
>>51493867
normism
of course if you were doing integer math you'd want an integer square root you fucking retard. otherwise you'd just use a square root function that takes a double or a float. fucking retard go hang yourself now.
>>51493894
>any two 63 bit or smaller integers
yes, and fails for integers larger than that
>"difference & 1" masks the least significant bit so it's 1 if difference is odd
sometimes
>>51493890
it's just how /g/ taught me how to program sempai
>>51493900
holy shit what is wrong with you
int mask(bool flag1, bool flag2, bool flag3)
{ return (((flag1 << 2) | (flag2 << 1)) | (flag3)); }
>IT RETURNS AN INT
>>51493925
cool story bro
>>51493663
>big dick playa
based C god
>>51493746
>overflow
>>51493944
>standard compliant story
ftfy
>>51493958
not a rounding error
return a long long instead
>>51493925
Note: "fails when most significant bit is set" means "fails for half of all integers".
(Or one quarter of integer pairs, given uniformly distributed integers)
>>51493941
>animoooooo
>>51493967
>compliant to 100% of target machines story
>>51493973
long long doesn't help with overflow
Wow this thread is cancer
>>51493996
long long average(int a, int b)
{ return (long long(a) + long long(b)); }
if long longs aren't twice as big as ints your compiler or target is shit
>>51494012
you're cancer
>>51493663
that's a common error that most fags do: thinking that a return value of EOF is equivalent to end of file or error; the shitweebs in here aren't even capable of understanding this
the point is the basic c integral types are shit and you should use fixed width integers
>>51494016
retard edition
long long average(int a, int b, bool* point5)
{
*point5 = (a%1) != (b%1);
return (a + b) / 2;
}
>>51493782
>or are you still only have limited skills as a developer?
This, which means I'm using Haskell.
>>51494016
>I don't know C
>my code doesn't even compile
looks like you're shit anon :)
how do you calculate the average of two ints ?
>"returning a double is autism REEEEEEEE " you should rather:
>use a third party library
>use obscure bitfiddling that doesn't take byteordering into account
>return a struct
>add the numbers and calculate the rest in shell
ITT: Web developers programming in C for the first time
>>51494046
>fixed width integers
less portable
>>51494061
C++ soz
long average(int a, int b)
{
return ((long) a + (long) b);
}
>>51494050
That'll overflow, then cast.
>>51494050
>a%1
you mean written by a retard
>>51494075
C is a meme, learn python instead.
>>51494088
yup use >>51494087 as well
>my solution worked best
>i got the most shit
>>51494090
&1 is not portable
>>51494093
Or you know, just return a double.
The sane way to do it.
>>51494087
you still don't know C. long an int are same size on a lot of architectures; it still overflows, you fucking moron
>>51494050
>modulo 1
Wtf are you doing retard?
>>51494102
>just return a double
not guaranteed to be big enough
>>51494083
Explain to me why I need integer types other than signed/unsigned n-bit integers and a single "pointer size" integer type.
>>51494075
>>use obscure bitfiddling that doesn't take byteordering into account
where did this take place?
Writing an Algorithm to create random Sudoku Puzzles.
It's pretty tough, but once I have that out of the way the rest of the program should be pretty easy (Mostly GUI stuff)
>>51494099
a % 2 == a & 1
a%1 returns 0 for all numbers
bool average(int a, int b)
{
return false; // all ints are special
}
>>51494083
oh hi again
>>51494134
>n-bit integers
you might not have them
>single "pointer size" integer type
pointers can have different sizes and might not be convertible to integers
>>51494149
typo, meant %2
>>51494149
>a % 2 == a & 1
false
>>51494099
>worked best
Long and int can have the same size. Long long however cannot.
>>51494180
>Long long however cannot
wrong
>>51494180
I used long long before but I wasn't sure if it was valid C or if it was just C++
but if your longs are the size of ints, your compiler or target is shit
who the fuck cares about averaging two integers, machines are dumb, just hire an indian for 1 USD.
>>51494201
fag
How difficult would it be to write a program that converts an .exe (Or something similar) back into its original Source Code?
NEW THREAD AT >>51494191
NEW THREAD AT >>51494191
>>51494166
still wrong, tard. do you even write C?
>>51494200
>your compiler or target is shit
ahhahahahahaha, you sheltered webfag; how's that css coming along? managed to look the same on 2 browsers? hahahahahah
>>51494200
>but if your longs are the size of ints, your compiler or target is shit
no argument here, but i'm pretty sure win32 has 32bit int and long.
>>51494179
It's correct. & 1 gives you the Least significant bit for all correct C implementations
>>51494210
It's not trivial, google reverse engineering.
int average (int a, int b)
{
return (a > b ? a : b);
// approximate result
// accuracy scales
// linearly with difference
}
>>51494200
it is more common that longs have size of int. long long being bigger
Discussions on the completely unreasonable goal of averaging two integers 100% portably in an O(1) fashion with no loss of precision below this line are PROHIBITED
----------------------------------------------------------------------------------------
>>51494237
if your implementation uses ones complement, the least significant bit won't be the "odd" bit for negative integers
>>51494237
>& 1 gives you the Least significant bit for all correct C implementations
yes, and the least significant bit tells you nothing about odd/even
>>51494253
for (int i = 0; i < a; i++)
result += (int) 1 / 2;
for (int i = 0; i < b; i++)
result += (int) 1 / 2;
//minimal rounding errors
>>51494237
% returns negative (or 0) for negative numbers. please stop posting, you are embarrassing yourself.
>>51494278
Which % 2 does as well.
>>51494294
No it doesn't
>>51494298
No. % 2 gives you the remainder.
>>51494257
Shit, you're right
>>51494319>cat a.c
#include<stdio.h>
int main(void)
{
printf("%d\n", -5%2);
}
>gcc a.c && ./a.out
-1
please, stop.
>>51494319
http://ideone.com/nOG9yb
>>51494334
>realized I'm dumb after shitposting 10 times about my "solution"
>>51494387
Cunt
>>51494382
inb4
>gcc
>>51494423
>(void)
>>51494423
>using compiler to prove c behavior
Die in a fire
Refer to standard or bust
>>51493763
>But they're not (unless you have an retarded implementation of fread/fwrite).
That's how they're implemented in newlib. Can't say I've looked at any other implementations.
>Not doing a syscall every byte anymore = no longer IO bound
>Then it starts to matter.
Assuming that fgetc/fputc are implemented as syscall wrappers is no less arbitrary than assuming fread/fwrite call the character functions in a loop.
>Doing an fread into a buffer is not "complicated code" anon.
It's more complicated than the code that was posted, and completely pointless. FILE streams are already buffered on all but the most obscure implementations.
>Yes, of course, but even if they're not, calling a function (or in this case, two functions) for every single byte should raise some eyebrows.
>Not necessarily, calling a function for every byte should be a huge red flag for any developer.
File IO is orders of magnitude more expensive than a function call. It doesn't matter.
>>51494165
>you might not have them
Good practice is to always choose the smallest integer type that meets the needs of your algorithm. If multiple sizes make sense depending on the use case, you use generics or multiple instances of the code. If you've done this, and the platform+compiler can't support your integer size, then it is because the algorithm can't run without modification on the platform, with your chosen compiler. You'll have to use bignums or something.
>pointers can have different sizes and might not be convertible to integers
Of course pointers can have different sizes, that's the point of having a "pointer sized" integer type. If you are on a platform that supports multiple, adjust the compiler flags to get something other than the largest supported size. Covering to/from integers is irrelevant. You use pointer-sized types for offsets into memory, and amounts of memory.
Having an integer type that isn't fixed, and isn't pointer sized, but is still platform-defined doesn't really make sense to me, except maybe as a hack to avoid needing generics.
>>51494454
it's implementation defined newfag.
https://en.wikipedia.org/wiki/Modulo_operation#Common_pitfalls
>>51494487
not in C. C99 saysa == (a/b) * b + a%b
>>51494466
>File IO is orders of magnitude more expensive than a function call. It doesn't matter
The point of buffering in the first place is that you do i/o in bouts at the time. What you are left with is a tight loop that moves bytes between two buffers in an insanely stupid way.
You wouldn't do write() one byte at the time, so why the fuck would you do memcpy one byte at the time?
>>51494487
Exactly, fag.
autism itt
>>51494628
but it is not. you both are retard. I am the superior programmer.
Check C99 standard 6.5.5.5-6
Why aren't you programming in c# or another fun language
>>51494666
>shills C#
>666
>>51494666
microsoft are the antichrist confirmed
>>51494603
I'm not convinced that there's anything stupid about it. It's clear, concise and obviously correct. These are good properties to have.
>>51494191
>>51494191
>>51494191
since other guy was a retard
>>51494532
so it's defined so that n % 2 will be -1 for odd n
>>51494408
>I'm mad now
>>51494479
>pointer-sized types
they might not exist, you fucking moron
>>51494213
>>51494860
No they weren't
