The h Bar

dmckee

22:00

I can try. But I'm a little rusty.

And someone "borrowed" my copy of the new testament and didn't give it back. Grrrr.

Bernardo Meurer

In my main I have a pointer to a struct that has elements h and w

0celouvsky

deus vult @dmckee

Bernardo Meurer

There, I can access these elements like

int foo = IST_logo->h;

IST_logo is the struct

Now, I have a function that takes SDL_Surface **_logo

How do I reference the element inside the function?

Balarka Sen

i wanted to make a religious joke about that but figured i'd just not say it

dmckee

Something like (*_logo)->h ought to do it, I think. Hard to be certain without looking at more of the code.

Bernardo Meurer

22:03

Ha

It's the parenthesis that were missing

interesting

dmckee

And I feel compelled to complain that you shouldn't be using _ prefixes. They are reserved.

Bernardo Meurer

I know, I will refactor the naming

I just need to bootstrap this

Thanks @dmckee!

0celouvsky

bootstrap?

dmckee

Lots of the precedence relationships for c's operators are weird enough that parentheses spring up in odd places.

@BalarkaSen In this instance "the new testament" is the 2nd edition of The C Programing Language by K&R.

Bernardo Meurer

@0celouvsky "Just make it work"

dmckee

22:08

Though someone needs to write an equivalently short and dense book for a newer version than c90. Alas Ritchie is gone.

Bernardo Meurer

Yeah, Ritchie is gone :(

I'm glad Knuth is still around

0celouvsky

why?

Bernardo Meurer

He's a Computer Scientist's Computer Scientist

0celouvsky

he's going to release another edition of his books

and they're gonna be more expensive than ever

@BernardoMeurer What does that even mean?

Bernardo Meurer

Yeah, he's nuts, he says he's going to volume 10

He's currently at 4

@0celouvsky Not sure, someone once said that about Grothendiek and I thought it sounded cool

0celouvsky

22:12

he was a nut

lots of these guys are bonkers

Bernardo Meurer

Perelman...

Balarka Sen

which guys

0celouvsky

> Although in applications we deal with analytic manifolds as a rule

wat

I've never once seen an analytic manifold used unironically

Lie groups don't count

Slereah

$\Bbb R^n$ is analytic

Hm

I think the guy who told me that maybe we could have contractible curves covered by simply connected charts if they were injective was wrong

Or... was he

I'm not sure

the example in mind is fairly convoluted

No I think he might be right

0celouvsky

@Slereah there are plenty analytic manifolds

But one never uses them as analytic manifolds

Slereah

22:29

wait what

The new coordinates don't include the old coordinates

Oh, I think those are just arrival times

nvm

I'll have to work out the coordinates myself

Bernardo Meurer

@dmckee Still here?

0celouvsky

23:10

I like when a book talks about things I already know

I can nod violently while reading

Slereah

Hm

dmckee

@BernardoMeurer Well, I'm back.

Slereah

If we have two spacetimes $g^1_{\mu\nu} = \eta_{\mu\nu} + k^1_{\mu\nu}$ and $g^2_{\mu\nu} = \eta_{\mu\nu} + k^2_{\mu\nu}$

Is $g_{\mu\nu} = \eta_{\mu\nu} + k^1_{\mu\nu}+ k^2_{\mu\nu}$ a spacetime, assuming that $k^1$ and $k^2$ have disjoint support

Bernardo Meurer

@dmckee How can I calloc a 2D array so that I can access it like foo[a][b]?

Slereah

Pretty sure it's correct

dmckee

23:16

Uh...

Bernardo Meurer

Yeah :P

dmckee

Try something like int *(foo[a]) = calloc(b,sizeof(int[a]));

Maybe.

Formally, of course it is a pointer, but you can still access it with array semantics.

Bernardo Meurer

@dmckee If I try that everything is red, lol

dmckee

But I have to warn you, that code using this kind of trickery is a pain to work with unless you can isolate all the clever bits.

Bernardo Meurer

"Compound array initializer is expected"

dmckee

23:20

@BernardoMeurer I take that is you IDE tell you it is systactically incorrect?

Bernardo Meurer

Yes

/home/meurer/src/prog/project01/main.c:75:5: error: variable-sized object may not be initialized
     int *(board[game_sz]) = calloc(game_sz, sizeof(int[game_sz]));
     ^~~

GCC doesn't like it either

dmckee

Then I'm stumped without getting a book and really poring over it.

I miss my K&R.

Bernardo Meurer

Hmmm

Let me try some pointer magic

DanielSank

Y U abuse compiler?

dmckee

^^ words that almost always presage disaster.

2

Bernardo Meurer

23:24

int **board = (int **)(calloc((size_t)(game_sz), sizeof(int *)));
for(int i = 0; i<game_sz; ++i) board[i] = (int *)calloc((size_t)(game_sz), sizeof(int));
board[0][0] = 1;

Works

@DanielSank Abuse? How?

DanielSank

3 mins ago, by Bernardo Meurer

Let me try some pointer magic

Bernardo Meurer

@DanielSank Oh, I didn't go for the magic :P

David Z

@BernardoMeurer makes sense, because to do multilevel indexing in C you need to have an array of pointers

Bernardo Meurer

I went for the verbose way

@DavidZ Yep :)

David Z

When I've had to do this I made a macro INDEX_2D(n, i, j) = j * n + i or something like that, then I could just do board[INDEX_2D(n, 0, 0)] = 1 and the like

I didn't want to bother with the complexity of arrays of pointers

Bernardo Meurer

23:27

@DavidZ Yeah I'm trying to steer away from macros

I think this pointer to pointer approach is nice enough

David Z

Sure, whatever works for you

I suppose you can get away with malloc-ing the first array because you're initializing all the elements right afterward anyway

Bernardo Meurer

This is already me refactoring my refactor to make it less refartorant so that the prof accepts my code

@DavidZ calloc is faster than malloc on linux :P

David Z

really? weird

DanielSank

@BernardoMeurer why?

Bernardo Meurer

Yep, for really interesting reasons, I have an article, one second

@DanielSank Because the kernel cheats

DanielSank

23:29

@BernardoMeurer Did you see what I was saying "why" to?

Bernardo Meurer

186

Q: Why malloc+memset is slower than calloc?

It's known that calloc is different than malloc in that it initializes the memory allocated. With calloc, the memory is set to zero. With malloc, the memory is not cleared. So in everyday work, I regard calloc as malloc+memset. Incidentally, for fun, I wrote the following code for a benchmark. ...

c malloc

@DanielSank Nope :P

Because I dislike macros

DanielSank

@BernardoMeurer o_O

Bernardo Meurer

I think they look ugly and are generally not needed

DanielSank

@BernardoMeurer Ho boy. Ok, try to write a nice RPC system.

Bernardo Meurer

I will using solely GOTO

and booleans

DanielSank

23:32

You'll get rekt.

@0celouvsky change your username to ILoveMeasureTheory.

David Z

@BernardoMeurer oh, that's different. malloc + memset will of course be slower than calloc. But you're not using memset or any equivalent, so that's not relevant for your code snippet.

Bernardo Meurer

@DavidZ My bad, I need everything to be 0-initialized

I also generally dislike malloc

I think it makes it easy to screw up and access unitialized memory

David Z

I'm just saying you don't need board to be zeroed in this case, not unless you're doing some multithreading nonsense or the like. Not that it really makes a huge difference.

DanielSank

Stop generally disliking stuff. Different stuffs are good for different things, and stuff.

Bernardo Meurer

@DavidZ Hm? I'm telling you I NEED it zeroed for my use case :P

@DanielSank I dislike GOTO

G. Bergeron

23:42

@BernardoMeurer Ever used one?

Bernardo Meurer

@G.Bergeron Every week

G. Bergeron

@BernardoMeurer why?

Bernardo Meurer

@G.Bergeron Coding in assembly for my MIPS CPU, Writing C code but the implementation of libc is incomplete (I'm writing my own kernel).

G. Bergeron

@BernardoMeurer In what physics coursework were you asked to do that o_0 ?

DanielSank

@G.Bergeron None.

Bernardo Meurer

23:46

None

DanielSank

Computer engineering courses.

@BernardoMeurer I beat you.

Bernardo Meurer

MIPS part for Computer Architecture

kernel part is to satisfy my masochistic tendencies

DanielSank

dafuq is MIPS?

Million Instructions Per Second?

Bernardo Meurer

An ISA

G. Bergeron

It's too low-level for me... I like reading about it, understanding it and using what others did!

Bernardo Meurer

23:47

Not very good one, but easy to make

G. Bergeron

Something about no interlocked pipeline

in the cpu

Bernardo Meurer

I'm the kind of Computer Limbo

G. Bergeron

@BernardoMeurer what do you mean?

Bernardo Meurer

I always go to the lowest level :P

G. Bergeron

@BernardoMeurer Me too! ... in theory.

@BernardoMeurer So when do you plan on refining silicon and doing lithography? :p

Bernardo Meurer

23:52

@G.Bergeron In a couple years :P

DanielSank

@BernardoMeurer False. I'm working with capacitors and circuit traces, boi.

rekt

Bernardo Meurer

I'm starting to work on rebuilding SH in a Spartan 6 right now

@DanielSank Yes, you also cannot compute anything that requires more than 15 bits

DanielSank

shaddap!

Bernardo Meurer

REKD

DanielSank

(I have never been rekt harder than that. Ever)

Bernardo Meurer

23:53

<3

That was meant as an incentive

Now you work not just for money and recognition, but for your honor to unrek yourself

G. Bergeron

@BernardoMeurer Hmm... that depends on how you count!

In a a way, you are on the digital level of abstraction and he is on the analog level (possibly). Hence, there are so much more degrees of freedom interacting with an analog computing power blasting away anything you're doing :p

Bernardo Meurer

@G.Bergeron I'll accept that when his computer runs Doom

Cause even my calculator does

G. Bergeron

@BernardoMeurer Bwah, your rigging the test in your favour

Bernardo Meurer

Am not! Running doom is a test standardized by the ISO

and recognized internationally

Transcript for