The Nineteenth Byte

3:00 PM

@Mego The one which mine was a duplicate of (and which has been suggested for yours too) is a popularity contest, which is frustrating if you've come up with an objective scoring method. However, at present I agree with Calvin's Hobbies second comment in the sandbox, so the scoring method isn't right yet

Anonymous

Rule #3 in my post disallows cat-like programs

Calvin's Hobbies

@Mego Yeah I missed that, but where is the line drawn between interesting and uninteresting?

trichoplax

@Mego I think that still just shifts the problem one place along. I can now make the minimum required attempt to golf, and enter that very short program

Calvin's Hobbies

Exactly

trichoplax

Ideally a scoring method should reward how well the code golfs a variety of different inputs, but that is very hard to judge without making it language specific

This is why the other one is still a popularity contest - we did try hard to think of a good winning criterion a year or so ago

Calvin's Hobbies

3:03 PM

@trichoplax e.g. Could one just replace \r\n with ; in javascript?

trichoplax

@Calvin'sHobbies I know there are a number of easy improvements to be made in python, and only one of them would be required to count as "not cat". In any language, the improvement doesn't have to be possible for a given input - that can never always be true anyway

Anonymous

Perhaps making the score the difference in bytes between your program and its output, when given itself as input? Except that would lead to just artifically inflating answers with whitespace/NOPs...

trichoplax

So just converting double spaces to single spaces would count, even though most code will be unchanged by it

@Mego Yes exactly - ideally it needs to be set test cases, but that can only work for a single language.

It is a very tough criterion to get right

Anonymous

How about the score is the sum of the difference in bytes for a few select inputs? Such as itself, a Hello World program, and a 99 Bottles program? Ideally the Hello World and 99 Bottles original programs would be chosen from a neutral source, like RosettaCode or esolangs

Calvin's Hobbies

People will still game that

One option is to define your own language and examples that are golfed. (with golfers still in any usual language)

Anonymous

3:08 PM

Maybe, but in the end, no matter how the challenge is stated, there will always be a way to game it.

trichoplax

For a single language it can work:

11

Q: Auto-meta-code-golf

You are sick of all of the codegolf challenges. Hence you decide to write a program that will automatically golf some Python code for you. There are 3 test cases: print quickSort([0,7,3,-1,8,10,57,2]) def quickSort(arr): less = [] pivotList = [] more = [] if len(arr) <= 1: ...

Anonymous

I wanted to make it as generic as possible, to allow as many languages as possible to compete. It'd be interesting to see what a CJam solution would look like.

trichoplax

I love the idea of a language agnostic version, so everyone can compete regardless of their chosen language, but the only way I can see it working is to have test cases for every language that someone wants to compete in...

Anonymous

I wouldn't mind putting in the legwork to create a test case for every language people submit answers for

Anonymous

Like I suggested before, the test cases would probably be the golfer itself, Hello World, and 99 Bottles

Calvin's Hobbies

3:14 PM

Could they optimize their answer after seeing the test cases?

trichoplax

You could suggest that in your sandbox post and see what people think. You could invite comments on your final posted question requesting new language test cases.

Dennis

@Mego Fixed the last test case. One last clarification: The questions will always be Is <Class> a <Class>? or Does <Class> have a <member>?, yes?

trichoplax

@Calvin'sHobbies Have you seen the third answer for the python golfing question...?

Anonymous

@Calvin'sHobbies I would prefer generic optimization that will perform better across all inputs, rather than optimizing simply to perform better on the test cases. But I don't see a reasonable way to do that, so there wouldn't be a restriction.

Anonymous

@Dennis Correct, those are the only two formats. You may safely assume that no incorrectly-formatted input will be given.

Dennis

3:19 PM

Great, thanks! I think I can start working on that explanation now.

Anonymous

Always glad to help

trichoplax

@Mego @Calvin'sHobbies if only the input is being golfed (not the answer itself) then hopefully tailoring to the test cases won't be too much of a problem - we just need test cases that cover most of the methods of golfing

Anonymous

Now how do I make a chat transcript?

Calvin's Hobbies

I'd think you'd need many more test cases than just Hello World and 99 bottles to get an accurate result.

trichoplax

@Mego The drop down arrow to the left

Calvin's Hobbies

3:20 PM

or scroll to the top and hit "full transcript"

Anonymous

Alright, updated the scoring rules. How does it look now?

Calvin's Hobbies

Esoland?

:P

Anonymous

You saw nothing

trichoplax

I wanna go!

Anonymous

All rides are Turing tarpits

Anonymous

3:32 PM

Brainfuck is much more literal than you'd be comfortable with

trichoplax

I think including the code as one of the test cases still leaves the loophole

Calvin's Hobbies

Yeah

If they just golf it anyway the difference will be 0

trichoplax

If you take that bit out, I'd see it as viable but controversial, and that's not a bad thing...

The only question I have is, what is the tie breaker between two programs in the same language that score the same?

Making it shortest code leads back to the same problems.

Simply first posted might work

Anonymous

Since the self-input score is only part of the total score now, and it's the difference in lengths, any gaming of the system will have minimal effect

Anonymous

I don't know about first posted, since I'm not disallowing optimizing via edits

trichoplax

3:35 PM

If I make my code N bytes longer, and all N of them are removed, I score N more, right?

Calvin's Hobbies

@Mego You'll have to deal with people answering in Hexagony and other crazy languages that probably have no 99 bottles implementations. Which means that you'll have to learn them, write the code, and decide what the ungolfed and golfed versions are.

trichoplax

@Mego Yes I should clarify that more. First to post that score, rather than basing on the initial answer time

Anonymous

@trichoplax True, that is a serious exploit still. Suggestions for a third neutral test case?

Anonymous

@Calvin'sHobbies When I said "collaboration with the community", I meant "write Hexagony code for me" :P

Calvin's Hobbies

But that's not fair

They will write the code to suit their own golfer then

trichoplax

3:37 PM

@Calvin'sHobbies That's a good point. Particularly as more test cases is better, but makes that problem even worse

Calvin's Hobbies

You have to ok it to make it impartial

trichoplax

It's OK for languages which dozens of people use, but a language that only one person here uses means they can hand pick their own test cases (unless you learn it too - as Calvin's Hobbies points out)

Anonymous

Yeah, I'll be the final judge on whether a test case is acceptable or not. I expect that people will mostly play fair when it comes to helping me create test cases for languages I'm not familiar with, for the sake of the contest

Anonymous

This is also a good excuse for me to learn the fundamentals of some new languages

Anonymous

Err, in the spirit of the contest*

Calvin's Hobbies

3:40 PM

@Mego I would not count on that. In my experience people usually are fair here, but they are also very vocal about making sure that no one else can be unfair.

Anonymous

@Calvin'sHobbies Hence why I am the final arbiter.

trichoplax

There are still likely to be people complaining that the test cases are not consistent between languages, and you will be making a lot of work for yourself, but this seems a lot closer to making it to main than before

Anonymous

There isn't a way to make it perfectly fair. The closest we can get is having a neutral party (i.e. me) make the call. I'm willing to put in the work, because this is an interesting question to me, and I'm eager to see what people come up with.

Calvin's Hobbies

@Mego But people won't like that either. Your ability to recognize golfability in languages you may have just learned is not an objective winning criterion.

Anonymous

True, but I don't currently see a way of making the challenge more fair.

Calvin's Hobbies

3:52 PM

One way is to restrict it to "common" languages that definitely have solutions elsewhere online.

But otherwise I'm just warning you that it's possible your question would be closed

Anonymous

Maybe restrict it to languages that have Hello World and 99 Bottles example code on Rosetta Code or Esolang?

Dennis

@Mego If I print 1s and 0s, does the output have to be one per line or can I just print them without separation? (It wouldn't save me any bytes, but it would make the truthy values consistent.)

Anonymous

@Dennis You should separate them in some way, whether it be spaces, newlines, or kittens. Outputing "110" is ambiguous - is that "1,1,0" (true, true, false) or "110" (true)?

Dennis

OK, I'll leave it as it is.

Calvin's Hobbies

4:13 PM

@Dennis :(

New Meta Posts

0

Q: Can I invent a language to solve a problem that's in the Sandbox?

We have rules about answering challenges with a language that was invented after the challenge was posted. Can I invent a language to easily solve a challenge that's currently in the Sandbox and then answer with that language once the challenge has been posted? I doubt this has happened for a w...

discussion answers sandbox

PhiNotPi

I've made some progress on my bitcoin challenge controller.

One thing I want to figure out, however, is how to scrape the bots from their answers.

To the very far right of this graph is a little line representing a random bot.

It outputs a random number each turn, losing money due to the 0.2% trading fee.

Anonymous

4:31 PM

codegolf.stackexchange.com/a/61124/45941 This answer is driving me nuts

Anonymous

I know it's undoable, but dealing with Java reflection is like herding hyperactive kittens

Dennis

@Mego It is done.

3

A: Golf Me An OOP!

CJam, 59 bytes q_'.e=\N/{)'?=\S/>_,(%}%/(__,*{(2$z~@f=.*\m*|}/ff{1$e=>:=N} This finishes instantly for both test cases. It either prints the second name name of the question or 1 (both truthy), or 0 (falsy). Try it online in the CJam interpreter: Chrome | Firefox Idea Because of the distin...

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

code-golf Objective Given a string in Markdown, determine how much of the string is in bold and italic. Test Cases **Bold** --> 0 *Italic* --> 0 **_BI_** --> 2 __*abcde*__ --> 5 This is in **bold and _this_ is in italic** --> 4 \**_BI_** --> 0 \\**_BI_** --> 2 This is not in ___ bold-italic _...

Anonymous

@Dennis Very nice, now we just need more answers :P

Anonymous

5:10 PM

Ahah, got the Java one

Transcript for