Language Development

Hello

3:17 AM

Hello @LeakyNun

@MDXF hi

what is the file triangular for?

Oops it's the executable

for what

Deleted now

Well when you make it generates the interpreter executable

I see

3:19 AM

I pushed without deleting / adding triangular to the .gitignore

I'll be making a PR in a few minutes

Okay, thanks

→ 7 messages moved to The C Programming Language

@LeakyNun Ping me when you PR

done

Oh :P

Thanks :-)

I'll be making another

but have you tested it before merging?

I may have a syntax error somewhere...

3:29 AM

Yep I tested it, nothing messed up

good

can you tell me what buf will be, if the code is abc?

Darn I had some code in there that would print buf based on a command-line argument

but I removed it for some reason

I can't run C anyway

I just mean theoretically

buf[0] = "a" and buf[1] = "bc"

        if (x < 0)
            x = 0;
        if (y < 0)
            y = 0;

then why isn't there a similar wrapper for x > y?

3:31 AM

Hm I never saw it being a problem since if buf[y][x] is 0 it will automatically terminate IIRC

And if x > y then buf[y][x] will be 0

but then why doesn't it terminate if you go west?

It does

If you go too far west

    if (buf[y][x] == 0)
        exit(EXIT_SUCCESS);

yeah I was right, I did put that catch in there

but if you go too far southwest, you won't terminate

1 min ago, by MD XF

    if (buf[y][x] == 0)
        exit(EXIT_SUCCESS);

buf is initialized to be filled with zeroes

        if (x < 0)
            x = 0;
        if (y < 0)
            y = 0;

if you go too far southwest, you'll be at [0][0], which would not be 0

if ((direction == 4 && x < 0) || (direction == 2 && y < 0) || (direction == 1 && (y < 0 || x < 0)))
    exit(EXIT_SUCCESS);

directions 4,2,1 are checked

3:35 AM

I need an enum for directions...

but southwest is direction 6

Wait no

nah you just need constants

I can make them for you

in my next PR

Southwest is direction 6, direction 6 increments y

increments y

So travelling southwest will never get to [0][0]

oh you're right

3:37 AM

Also I've got 50 test programs on my offline PC and I'll test all those late tonight with the new code so if there are any bugs in the code they won't go unnoticed for more than 12 hours

ok

right, I just found a bug.

caused by me.

Oh?

I thought the directions were arranged in a circular order

They are...

wait that's not circular

duh

if (x < 0)
    x = 0;
if (y < 0)
    y = 0;

what's this code for?

3:41 AM

Error catching

'cause if either of those are less than zero it'll segfault

but they can't be

you already detected them beforehand

Yeah I think that's either a safety thing or a fallthrough from an old version

Don't remove it, I'll remove it tonight when I'm checking everything

  case '|': direction = SOUTH; break;

but in your docs it should be v

Where do I have that??

Is that really in there? *smacks head*

I changed the 7 to SOUTH

3:44 AM

Ha well looks like I don't have one test case

No I know SOUTH = 7 but I was talking about the case '|'

I guess I never bothered to test v

Ping when PR

Also if you don't use a computer on which you can test C, I'd recommend using cloud9, it's what I use to pull/test C repos

@MDXF PR

<insert a bunch of const ints here> -> enum { UNDEF, NORTHWEST, NORTH, NORTHEAST, WEST, EAST, SOUTHWEST, SOUTH, SOUTHEAST};

Why introduce so much confusion with dx and dy?

    switch (direction) {
      case NORTHWEST:   dx = -1; dy = -1; break;
      case NORTH:       dx = -1; dy =  2; break;
      case NORTHEAST:   dx =  0; dy = -1; break;
      case WEST:        dx = -1; dy =  0; break;
      case EAST:        dx =  1; dy =  0; break;
      case SOUTHWEST:   dx =  0; dy =  1; break;
      case SOUTH:       dx =  1; dy =  2; break;
      case SOUTHEAST:   dx =  1; dy =  1; break;
    }

    x += dx * (skip + 1);
    y += dy * (skip + 1);

That's a bit less manageable than it used to be...

4:03 AM

because I don't like how the skip works

like, seriously? You skip it one by one?

Well for one thing, the IP still needs to move if skip is nonzero

skip is not always true/false so x += dx * (skip + 1); is gonna really bork stuff up

@MDXF I know, which is why I put +1

@MDXF can you show me how?

skip may be any value

go on

Oh you broke that part :P

github.com/aaronryank/triangular/blob/master/triangular.c#L99

skip = (size && stack[size-1]);

skip needs to be stack[size-1], not stack[size-1] == true

4:07 AM

wait a sec, fixing

what... how come 1&&0 is 0, 1&&1 is 1, and 1&&2 is 1?

@LeakyNun Because 2 evaluates to true

&& != ==

&& just checks if both values are nonzero

I thought && is like and in Python lol

Nope haha

    skip = skip < 0 ? 0 : skip;

Are you ok with this line?

@MDXF

When will skip ever be ... good point, that's a bug. Yeah that looks okay, where're you putting it?

4:15 AM

just before

    x += dx * (skip + 1);
    y += dy * (skip + 1);

...I thought we were removing those lines

@MDXF PR fixed

@MDXF I intend those lines are not borked

and are equivalent with what you originally have

and are faster when skip = 10000

Yes but say dx = 1 and skip = 3. Won't x += dx * (skip + 1); evaluate to x += 1 * 4?

And we don't want x incremented by 4

Hmm... putchar(size ? stack[size-1] : 0); is not good, we don't want to print null bytes...

@MDXF what would x be incremented by instead?

@MDXF so just don't print anything?

Wait a sec

Oh it's all making sense now

Does x += dx * (skip + 1); automatically increment x to however many it should be if skip is not 0 or 1?

Like if skip = 4 then it'll skip the correct number of commands. Ohhh

And skip = 0; Yup yup looks great, sorry about that

4:22 AM

@MDXF yes

My bad, I'm tired :P

@MDXF so what do we do here?

I'm still looking over the PR @LeakyNun

@LeakyNun I don't think these safety nets are necessary; if size < 1 then the compiler will just default stack[size-1] to 0

I thought stack[-1] is segfault

@LeakyNun I'm pretty sure it's not but I'll test it

@LeakyNun Nope, defaults to 0

4:30 AM

@MDXF then what is this?

@LeakyNun Hm, I think it was my best guess as to why I put that useless code in there. Actually, I'm pretty sure that code was to prevent x or y going under 0, and then the error-check that exits if the code is empty later ... yeah never mind

and I still don't like stack[-1] being called...

It was useless

@LeakyNun Well neither do I...

but this

skip = (size ? stack[size-1] : 0) < 1;

If size isn't 0 then it just compares 0 < 1

Which is just a waste of the compiler's time

Maybe skip = size ? (stack[size-1] : 0 < 1) : 0

Actually I can just do that @LeakyNun do you have any more changes to make?

I'm going to merge your PR

@MDXF it's actually UB

@LeakyNun GCC/TCC/Clang (the only compilers you should use) really handle UB well

4:36 AM

my conjecture is that if you do x[-1000] eventually you access non-zero addresses

@LeakyNun Okay yeah that seg-faults

But it will never get that low

@MDXF there's an obvious syntax error in your code

@LeakyNun Where?

Not in my code... I already tested it

follow my reply arrow

This? skip = size ? (stack[size-1] : 0 < 1) : 0

4:39 AM

yes

Oh duh

skip = size ? (stack[size-1] < 1) : 0

Thanks

Anyway I'll implement a few things; can I merge your PR now?

@LeakyNun

ok

@LeakyNun But still it will never get that low

        if (buf[y][x] == 0)
            exit(EXIT_SUCCESS);

So y and x will never get below -1

This too

if ((direction == WEST && x < 0) || (direction == NORTH && y < 0) || (direction == NORTHWEST && (y < 0 || x < 0)))
    exit(EXIT_SUCCESS);

@MDXF this is redundant, it wont get that low because it wont even be executed

and for other cases, it's your language, so you use UB at your discretion

Either way I'm removing the UB

Well, you removed the UB

Also for some reason your eight consts in the switch didn't work

but I fixed that with an enum

4:48 AM

cool

I have no idea why those didn't work

triangular.c:73:13: error: case label does not reduce to an integer constant

maybe not global?

I think they were global

Ah well, I'll look into that some other time

I gtg in 6 minutes @LeakyNun so if there's anything else you'd like to bring up...?

@MDXF I'll just create 1000 PR's

you can discuss those with me when you come back

Nice :P I may make changes over"night" (depending on your timezone) but they won't be pushed 'till "tomorrow" morning, and they'll just be bugfixes

4:54 AM

how many hours later?

Oh crap, I'm out of town tomorrow, just remembered

alright

I may not have time to do anything on GitHub until ... say in 20 hours, sorry

no problem

In that case I won't do anything overnight aside from tests. Thanks for the help

5:49 AM

@MDXF you write ] move the instruction pointer back to the most recent ( if the ToS is > 0, but you only check stack[size-1]: what is the ToS is negative?

7:39 AM

@MDXF UB much

2:09 PM

@LeakyNun Then it shouldn't do anything; in Triangular, negative numbers act as falsy values

@MDXF yes, but in your code it doesn't.

your code checked for truthiness.

Oh crud

I'll check on that

I thought you are not available until 11 hours later

Neither did I but my flight got cancelled

oh...

my condolences

2:11 PM

¯\_(ツ)_/¯

Fixed ]

did you check my code before merging it?

Yep

alright

There was one bug but I noticed it overnight and fixed it a few minutes ago

If the direction was North, dy was getting set to 2 instead of -2

8:22 PM

@LeakyNun I figured out why having a const int in a switch case didn't work, see this

9:15 PM

@MDXF I think triangular may not be TC

@WheatWizard how so?

There isn't any arbitrary stack acess

What do you mean, like printing the non-top stack value?

You can't perform an action on the second value without destroying the top value

You can perform =

9:16 PM

this doesn't necessarily mean its not TC, but its usually a sign

Thats not what I mean

A program that adds the 2nd item to the 3rd is impossible

Ohh yeah I meant to implement something like I did in Decimal

Gimme ten minutes

@DJMcMayhem Could you move these recent messages from me and WheatWizard to this room?

If it is TC its going to have to have a really strange memory representation

@WheatWizard what do you mean?

I might be able to make a quine anyway though

For example there is a subset of Brain-Flak called Microflak

Micro flak is technically turing complete,

Link?

9:20 PM

there is no link

Its all on paper at the moment

Oh, you're writing it. ok

But the point is that Microflak is TC, but all data has to bre represented in binary because it can't distinguish between non zero numbers

That seems like a very complex language; I'm excited to see it

So if you want to say add two numbers, you have to pass the numbers in a really esoterric format, in order for the language to compute any thing

Yeah. Well, Triangular won't be that way, I'm adding "memory" like in Decimal

Decimal has one memory value and a stack

9:23 PM

Its Brainflak without [...],<...>, (), or[]

How is that even possible?

Really esoterric input formats

What's a good name for a command that takes a value from a stack and puts it somewhere else without popping it?

stash?

sounds good

Like leave it on the board?

DJMcMayhem

@MDXF I'll have to switch to my computer, just a sec

9:25 PM

Yeah

how do you pick it back up?

For example, say we have two stacks: I need a name for a command to transfer the top of one stack to the other

@DJMcMayhem Okay, thanks

DJMcMayhem

Where do you want me to start?

Without popping it

Ah like Klein's (...) ?

9:26 PM

I guess so; never used Klein

ah ok

pull?

Yes that works

Maybe here? but if they don't need to be moved then don't bother

DJMcMayhem

← 18 messages moved from The Nineteenth Byte

Okay, three commands to interface with the second stack (which is currently not even a stack but just a single memory value): Pop, pUll and Stash

How does one compile triangular?

9:30 PM

make

Should that be in the README? Probably

Ah i didn't see the makefile

Does triangular strip whitespace like hexagony?

Yep

"Source code is read in the shape of the smallest triangle that the length of the code can fit into. Whitespace is ignored."

(emphasis mine) (text also mine lol)

       ,
      1 ,
     2 . .
    3 . . .
   4 . . . .
  5 . . . . .
 > . ( .%! / )
. > ( . . . . ]

What should that do?

No way, that's a quine??

Oh wait nvm haha

No its not

:p

9:33 PM

It will push the values 1 2 3 4 5 to the stack, direct the IP east, uh... I think that's a bit messed up

> . ( .%! / )?

ah I see

thanks

Remove the . directly to the right of the (

Yeah

Changes pushed, it should be more TC now

P to pop ToS and store the popped value in memory, S to stash the ToS in memory without popping it, U to pull the memory onto stack

Should make stuff easier

           .
          \ .
         1 . .
        2 . . .
       3 . . . .
      4 . . . . .
     5 . . . . . .
    . . . . . . . .
   . . . . . . . . .
  / . . . . . . . . <
 . . p ~ * * 9 2 5 < .
. . . . . . . . . . . .

What should that do?

shouldn't it print Z ?

What is the 1 2 3 4 5 doing? it's never reached

Wait why not?

9:41 PM

Wait nvm

whoops, sorry

I forgot the default IP moves by default lol

Just a second...

It seems to get stuck in an infinite loop, with no output

That's... strange

Looks like a bug

Try using the command-line options --verbose --display-code in that order

verbose will prettyprint the triangle, display-code will display the code as it reads it

            .
           \ .
          1 . .
         2 . . .
        3 . . . .
       4 . . . . .
      5 . . . . . .
     . . . . . . . .
    . . . . . . . . .
   / . . . . . . . . <
  . p ~ i * * 9 2 5 < .
 . . . . . . . . . . . .

.........<......../..54321\........<529**i~

9:45 PM

so it does this: .........<......../..54321\........<529**~

hm....

very strange

It doesn't infinite-loop for me...

it reads a character and then exits

Oh

I didn't see that

same for me

~ reads a character

Oh you thought that was an infinite loop

yeah

9:46 PM

Makes sense; that's more/less why I added --display-code

I'm dumb :p

Common mistake, np :P

if I change ~ to an @ I don't get any output sitll

Hm just a min

It prints a Z for me

hm

9:48 PM

try turning off --display-code :P

Its not on

And you don't see anything?

What character set is your terminal in?

character set?

ASCII, EBCDIC, etc

ASCII

9:49 PM

This is my --display-code: .........<......../..54321\........<529**@Zp..

the Z being the output

Huh then it should work

hm maybe I magically fixed a bug while not trying to fix a bug; pull maybe?

my display code is empty?

@WheatWizard It is??

I'm up to date

Hm

Is your display-code really empty?

Oh wait I forgot --verbose

9:51 PM

Yeah sorry that's something I really need to fix

My display is the same

.........<......../..54321\........<529**@Zp..

Then it's working right

is it?

I'm not getting output

The Z is the output

but only when I put --display-code on?

9:53 PM

I should use ANSI colors to make it possible to differentiate the output from the code

That's... quite strange

I get the same output without display-code, and display-code doesn't change anything in the interpreter

Ok well I'm going to try and ignore it

What system are you using?

OSX

My C compiler is super old if it makes any difference

Well for one thing OSX is trash... and what compiler?

gcc

9:55 PM

Hm... version 4 or higher?

yes

Its not that old

Oh hey

it works in terminal

Yeah no difference when using gcc-4.8 for me

just not iTerm

Oh, what were you doing it in?

Oh

thats weird

9:57 PM

Yeah. Freaking Apple...

actually it works in iTerm

just not fish

Never used fish

very nice

much better than bash imo

I prefer cmder for Windows and Zsh for Linux

I prefer fish for both OSX and Linux

9:59 PM

Speaking of fish, is ><> 2-dimensional?