Coding Horrors: You can gather all your technical debt into one easy framework!

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Coding Horrors: You can gather all your technical debt into one easy framework!

«‹›1503 »

Catalyst-proof: May 11, 2011; better waste some time with you

Internet Janitor posted:

For comparison, here's somebody else's C implementation based on the spec:
https://github.com/swetland/dcpu16/blob/master/dcpu.c

Am I allowed to think this is very elegant code?

# ? Apr 4, 2012 20:01

Adbot: ADBOT LOVES YOU

# ? May 21, 2024 02:56

mjau: Aug 8, 2008

Internet Janitor posted:

For comparison, here's somebody else's C implementation based on the spec:
https://github.com/swetland/dcpu16/blob/master/dcpu.c

That's not the same thing though. That screenshot from before was definitely a 6502 emulator (even the opcodes match), this dcpu thing is completely different. It's got different registers, all of which are 16-bit, opcodes and ops are different (has stuff like multiply and divide but missing other things), etc.

mjau fucked around with this message at 20:33 on Apr 4, 2012

# ? Apr 4, 2012 20:26

Jabor: Jul 16, 2010; #1 Loser at SpaceChem

Threaded code? Pffft. Should be writing a jiter to get some actual performance or something, I mean I don't see any timing data or anything to indicate he's going the cycle-accurate route.

# ? Apr 4, 2012 21:12

1337JiveTurkey: Feb 17, 2005

Jabor posted:

Threaded code? Pffft. Should be writing a jiter to get some actual performance or something, I mean I don't see any timing data or anything to indicate he's going the cycle-accurate route.

I think I remember someone that did an emulator which just translated the instructions to JVM opcodes but it seems like it'd be a bitch to actually split everything into procedures when developers historically could and did just screw with the stack.

# ? Apr 4, 2012 23:41

tef: May 30, 2004; -> some l-system crap ->

welp a game designer writes terrible code

# ? Apr 5, 2012 01:31

That Turkey Story: Mar 30, 2003

tef posted:

welp a game designer writes terrible code

I didn't know you were a game designer!

# ? Apr 5, 2012 01:39

Internet Janitor: May 17, 2008; "That isn't the appropriate trash receptacle."

mjau posted:

(has stuff like multiply and divide but missing other things)

My BCD arithmetic mode! :negative:

But yeah, I should've recognized the 6502 register set in his locals panel in that screenshot. Whoops.

# ? Apr 5, 2012 02:51

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

1337JiveTurkey posted:

I think I remember someone that did an emulator which just translated the instructions to JVM opcodes but it seems like it'd be a bitch to actually split everything into procedures when developers historically could and did just screw with the stack.

I doubt that ever could have worked. JVM performs validation on every branch to make sure that local types and stack lengths/types match up as well. You would have to out-clever the original developers by padding stacks and knowing when to pop.

# ? Apr 5, 2012 02:58

PDP-1: Oct 12, 2004; It's a beautiful day in the neighborhood.

1337JiveTurkey posted:

I think I remember someone that did an emulator which just translated the instructions to JVM opcodes but it seems like it'd be a bitch to actually split everything into procedures when developers historically could and did just screw with the stack.

They didn't just screw with the stack, they also used lower-address portions of the stack space to do scratch calculations on the assumption that those memory locations were probably not holding valid stack data. Considering that things like the 6502 had a 256 byte stack that must have been tricky.

To sperg on about 6502 stuff: If you read through the ROM source code for the Apple II, you'll find comments where Wozniak divided his team into 'perverts', meaning software guys, and 'non-perverts' meaning hardware guys. He would lock off sections of the hardware interop code with comments like "Perverts may not pervert the following code for any reason" and then end the section with a comment "Perverts may pervert the following code for any purpose". Before there was Subversion, they used Perversion as source control.

Internet Janitor posted:

My BCD arithmetic mode!

The very concept behind BCD is a coding horror that belongs in this thread.

# ? Apr 5, 2012 03:38

Jabor: Jul 16, 2010; #1 Loser at SpaceChem

PDP-1 posted:

The very concept behind BCD is a coding horror that belongs in this thread.

How so? A BCD encoding makes it far more efficient to format a number for display, at the cost of some additional complexity in performing arithmetic on it. It's the correct tradeoff to make in some situations.

# ? Apr 5, 2012 04:06

Thermopyle: Jul 1, 2003; ...the stupid are cocksure while the intelligent are full of doubt. �Bertrand Russell

I like how almost every time someone posts a horror someone else knows why it's not a horror or not as horrible as it first seems.

I think it means there are no horrible programmers.

# ? Apr 5, 2012 04:21

evensevenone: May 12, 2001; Glass is a solid.

mjau posted:

That's not the same thing though. That screenshot from before was definitely a 6502 emulator (even the opcodes match), this dcpu thing is completely different. It's got different registers, all of which are 16-bit, opcodes and ops are different (has stuff like multiply and divide but missing other things), etc.

the "DCPU" is the cpu you can program in his new game, which is apparently basically a 16-bit 6502 clone. So I think they are the same thing.

# ? Apr 5, 2012 05:40

Senso: Nov 4, 2005; Always working

And now there's already a JS web-based version.

# ? Apr 5, 2012 07:31

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

evensevenone posted:

the "DCPU" is the cpu you can program in his new game, which is apparently basically a 16-bit 6502 clone. So I think they are the same thing.

In no way is this a 6502. Note that Notch originally wrote a 6502 emulator to see if that would be viable for his game (which is what's featured in the screenshot), but realized that the 6502 both provided a lot of functionality that he didn't want, and didn't provide functionality that he wanted.

# ? Apr 5, 2012 07:42

1337JiveTurkey: Feb 17, 2005

Suspicious Dish posted:

I doubt that ever could have worked. JVM performs validation on every branch to make sure that local types and stack lengths/types match up as well. You would have to out-clever the original developers by padding stacks and knowing when to pop.

For very early systems I doubt that even if someone did get something working it'd be faster than interpretation due to all the edge cases that have to be tackled. Less limited systems seem more amenable because there's less pressure to take extreme measures for performance and some particularly troublesome options become impossible. For example self-modifying code in an architecture with a split cache and jumping into the middle of an instruction with a RISC instruction set.

I did find a project called JPSX which according to reports demonstrated full speed emulation of the Playstation using byte code translation with software graphics rendering to boot. There's a Google Code page for it with absolutely nothing though.

# ? Apr 5, 2012 08:57

tef: May 30, 2004; -> some l-system crap ->

code:


from StringIO import StringIO
from httplib import HTTPResponse
class FakeSocket(StringIO):
    def makefile(self, *args, **kw):
        return self

def httpparse(fp):
    socket = FakeSocket(fp.read()+"\r\n")
    response = HTTPResponse(socket)
    response.begin()

    return response

# ? Apr 5, 2012 11:57

trex eaterofcadrs: Jun 17, 2005; My lack of understanding is only exceeded by my lack of concern.

Thermopyle posted:

I think it means there are no horrible programmers.

I think it means we're all horrible programmers.

# ? Apr 5, 2012 13:02

Beef: Jul 26, 2004

Internet Janitor posted:

For comparison, here's somebody else's C implementation based on the spec:
https://github.com/swetland/dcpu16/blob/master/dcpu.c

Broken link, it can now be found on:
https://github.com/swetland/dcpu16/blob/master/emulator.c

Much nicer and probably faster than the notch botch.

edit: spoke too soon:

code:

if ((op & 0xF) == 0) {
		switch ((op >> 4) & 0x3F) {
		case 0x01:
			a = *dcpu_opr(d, op >> 10);
			if (d->skip) {
				d->skip = 0;
			} else {
				d->m[--(d->sp)] = d->pc;
				d->pc = a;
			}
			return;
		default:
			fprintf(stderr, "< ILLEGAL OPCODE >\n");
			exit(0);
		}
	}

# ? Apr 5, 2012 16:11

Look Around You: Jan 19, 2009

Beef posted:

Broken link, it can now be found on:
https://github.com/swetland/dcpu16/blob/master/emulator.c

Much nicer and probably faster than the notch botch.

edit: spoke too soon:
code:
if ((op & 0xF) == 0) {
		switch ((op >> 4) & 0x3F) {
		case 0x01:
			a = *dcpu_opr(d, op >> 10);
			if (d->skip) {
				d->skip = 0;
			} else {
				d->m[--(d->sp)] = d->pc;
				d->pc = a;
			}
			return;
		default:
			fprintf(stderr, "< ILLEGAL OPCODE >\n");
			exit(0);
		}
	}

Maybe it's just me but that looks like it sticks to the semantics of the (fictional) ISA fairly well. That code is implementing the JSR instruction, and it's fairly readable I think. Using a switch statement for one case is odd, but other than that it doesn't look terrible for interpreting machine code. Actually, on thinking about it, the switch may be there for adding future "non-basic" instructions, as JSR is the only one specified right now.

# ? Apr 5, 2012 16:37

Captain Capacitor: Jan 21, 2008; The code you say?

tef posted:

code:


from StringIO import StringIO
from httplib import HTTPResponse
class FakeSocket(StringIO):
    def makefile(self, *args, **kw):
        return self

def httpparse(fp):
    socket = FakeSocket(fp.read()+"\r\n")
    response = HTTPResponse(socket)
    response.begin()

    return response

The only way this makes sense is if you need to mock up a third party's response during unit tests or something. At least that's the only way my brain will let me comprehend it.

VVVV Just read through httplib.py and now I see what you mean.

Captain Capacitor fucked around with this message at 19:20 on Apr 5, 2012

# ? Apr 5, 2012 17:05

tef: May 30, 2004; -> some l-system crap ->

It's what happens when you try to parse http messages 'offline' from files. Nearly every parser is intertwined with socket code.

# ? Apr 5, 2012 19:02

Captain Capacitor: Jan 21, 2008; The code you say?

There's always this if you can use it.

# ? Apr 5, 2012 19:22

abiogenesis: Feb 4, 2009

code:

public <CT extends IC, DT extends ID<? super CT>> foo (ITT<CT> cClass, ITT<DT> dClass , List <? super DT> dList, List<? super aClass> otherDList)

This is government code.

This code is protecting your borders.

This is only one line of a 8000 line file containing similar methods :cry:

# ? Apr 5, 2012 20:03

CGameProgrammer: Nov 5, 2008

baquerd posted:

Any of these are fine:

code:

if (thing) doStuff();

if (thing) {
  doStuff();
}

if (thing) 
{
  doStuff();
}

if (thing) { doStuff(); }

This is terrible:

code:

if (thing)
  doStuff();

I think a good compromise is:

code:

if (thing)
{ doStuff(); }

# ? Apr 6, 2012 10:04

fishbacon: Nov 4, 2009; wonderful yet strange smell

CGameProgrammer posted:

I think a good compromise is:
code:
if (thing)
{ doStuff(); }

What about?

code:

if(thing)
{ 
   doStuff(
   ); 
}

Personally I like

code:

if (thing){ 
   doStuff(); 
}

best.

fishbacon fucked around with this message at 10:11 on Apr 6, 2012

# ? Apr 6, 2012 10:09

tef: May 30, 2004; -> some l-system crap ->

Captain Capacitor posted:

There's always this if you can use it.

The raw object in requests points to the underlying urlib3 structures. Which in turn wrap httplib. Which has socket code entangled with the parser.

There are two parsers afaik that aren't intertwined with socket code

http-parser - contains a wrapper around nginx c code & a pure python parser

the pure python one doesn't seem to be rfc compliant or worse, reality compliant (specifically some servers send LF only and not CRLF, and a few get defate/gzip mixed up, head responses don't have a body but will have a content-length, and 204, 304 will not contain a body but can contain a length parameter)

and the http parser I ended up writing that has been battle worn with some terrible websites

edit: and sometimes you get crazy loving http messages - an earlier version of wget with warc support, decided to preserve the headers but unwrap the body, so you have to be able to skip specific headers when parsing.

tef fucked around with this message at 10:29 on Apr 6, 2012

# ? Apr 6, 2012 10:22

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

PHP's == will convert arguments to floats if they look like numbers, even if both arguments are strings. That site has a lot of PHP WTFs, but this is the first time I've seen this one and I'm floored.

# ? Apr 7, 2012 01:31

McGlockenshire: Dec 16, 2005; GOLLOCKS!

I see we're back to the PHP part of the spiral...

That WTF is actually worse than it looks at first glance. See, it's not silently converting the string to a float, it's converting the string to an integer, failing to do that because the number is too large to store as an integer, then converting it to a float.

# ? Apr 7, 2012 02:53

csammis: Aug 26, 2003; Mental Institution

Hahaha, the PHP Sadness guy turned off commenting on his site. I guess too many people called out too many of his whines. I remember one where it was pretty clear he'd never understood what static meant.

e: to be clear about what I am saying: there is much to be sad about in PHP. This guy just couldn't be told that his complaints were themselves horrors.

# ? Apr 7, 2012 03:44

shrughes: Oct 11, 2008; (call/cc call/cc)

http://www.codinghorror.com/blog/2012/04/speed-hashing.html

codinghorror posted:

Hashes are designed to be slow

codinghorror posted:

In reality the usable space is substantially less; you can start seeing significant collisions once you've filled half the space, but half of an impossibly large number is still impossibly large.

# ? Apr 7, 2012 05:27

Hammerite: Mar 9, 2007; And you don't remember what I said here, either, but it was pompous and stupid.; Jade Ear Joe

shrughes posted:

http://www.codinghorror.com/blog/2012/04/speed-hashing.html

I am struggling to understand why he describes checksums as being a way of assuring "uniqueness", or why he describes them as being like somebody's full name. The older article linked to in his second paragraph demonstrates that he does understand the purpose of a checksum, but the comparison to people's names seems fatuous.

# ? Apr 7, 2012 05:54

het: Nov 14, 2002; A dark black past
is my most valued
possession

shrughes posted:

http://www.codinghorror.com/blog/2012/04/speed-hashing.html

codinghorror posted:

If Moore's Law anticipates a doubling of computing power every 18 months, that's like peeking 10 years into the future. Pretty amazing stuff, isn't it?

Oh come on

# ? Apr 7, 2012 05:58

Opinion Haver: Apr 9, 2007

It's funny because he has a good point at the end: use long passwords for yourself and use bcrypt or something for your password hashes. It's just cloaked in the usual Atwood bullshit.

# ? Apr 7, 2012 06:08

Blotto Skorzany: Nov 7, 2008; He's a PSoC, loose and runnin'
came the whisper from each lip
And he's here to do some business with
the bad ADC on his chip
bad ADC on his chiiiiip

Jeff Atwood has been publicly struggling with passwords for several years, and with pontificating on subjects of which he has little understanding for his tenure as a blogger

# ? Apr 7, 2012 06:10

Huragok: Sep 14, 2011

Otto Skorzeny posted:

Jeff Atwood has been publicly struggling with passwords for several years, and with pontificating on subjects of which he has little understanding for his tenure as a blogger

Dog barks, news at 11

# ? Apr 7, 2012 06:16

Hammerite: Mar 9, 2007; And you don't remember what I said here, either, but it was pompous and stupid.; Jade Ear Joe

Het posted:

Oh come on

That doesn't seem too far off. I get 10.84 years using his numbers, so he's in the right ballpark.

# ? Apr 7, 2012 06:38

Bruegels Fuckbooks: Sep 14, 2004; Now, listen - I know the two of you are very different from each other in a lot of ways, but you have to understand that as far as Grandpa's concerned, you're both pieces of shit! Yeah. I can prove it mathematically.

Hammerite posted:

That doesn't seem too far off. I get 10.84 years using his numbers, so he's in the right ballpark.

That would be along the lines of saying "between the ages of zero and nineteen, I never scored. But one time when I was twenty, my friend had sex (with a woman). That is infinitely more sex than I ever have had. My mom was wrong. This is the future."

# ? Apr 7, 2012 09:58

tef: May 30, 2004; -> some l-system crap ->

Oh jeff

# ? Apr 7, 2012 12:55

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

hieronymus posted:

That would be along the lines of saying "between the ages of zero and nineteen, I never scored. But one time when I was twenty, my friend had sex (with a woman). That is infinitely more sex than I ever have had. My mom was wrong. This is the future."

What? He's clearly giving an analogy to those who aren't familiar with the power of GPGPUs for this sort of processing: "this is how powerful your CPU will be in ten years"

# ? Apr 7, 2012 16:34

Adbot: ADBOT LOVES YOU

# ? May 21, 2024 02:56

Blotto Skorzany: Nov 7, 2008; He's a PSoC, loose and runnin'
came the whisper from each lip
And he's here to do some business with
the bad ADC on his chip
bad ADC on his chiiiiip

Suspicious Dish posted:

What? He's clearly giving an analogy to those who aren't familiar with the power of GPGPUs for this sort of processing: "this is how powerful your CPU will be in ten years"

Incidentally, it isn't. The corollary (really a misinterpretation but whatever) of Moore's Law has been dead for years, and now Dennard scaling is gone too

# ? Apr 7, 2012 18:40

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Coding Horrors: You can gather all your technical debt into one easy framework!

«‹›1503 »