@DJMcMayhem Thanks for moving that discussion to chat. I wanted to, but saw no option to do so. Is there a way I could have done it myself?

@Deadcode I think after a certain number of comments (3-4 each?) there's an option to create a chat room, but there's no way for regular users to move old comments into chat. (Because it would effectively allow you to delete other users comments).

But you can always flag requesting one of us to move a conversation to chat.

@DJMcMayhem I think it's when two people have posted at least 3 undeleted comments each

@DJMcMayhem I've seen that option to create a chat room appear once, but in this case it never appeared.

@EriktheOutgolfer That would have come up here

@Deadcode BTW, awesome answer. I was going to bounty it, before I noticed that you already had a bounty up, so I guess I'll have to wait ;)

@DJMcMayhem the countless times I've faced a situation where 5 people have each posted 2 comments but no the button won't appear...

@DJMcMayhem Thank you very much :D

@DJMcMayhem Okay, I awarded my bounty to someone :)

BTW, I would love for someone to write a PCRE or .NET regex for matching factorials. I'm just beginning to learn to exploit those (most of my focus has been on learning to work around not having them) so I probably wouldn't come up with a very well-golfed answer myself.

er, to exploit forward/nested backreferences I mean

@primo @H.PWiz @JoKing

(PCRE, Perl, or .NET, that is)

oh and of course @MartinEnder though he's not here.

@jimmy23013 also.

useless pings are useless...

Useless in that the site won't notify people who aren't in the channel, or in that those people are way too busy to respond to a ping such as that?

It didn't autocomplete them, which seems to be a hint.

yes, they don't work

they've not been here in the last 7 days

@EriktheOutgolfer this autocompleted for me o.O

it doesn't

what do you mean?

or do you mean specifically the people he pinged

yes his last two pings

ah gotcha

Have there been challenges where the question explicitly says people answering can choose either to optimize for size or speed? Or is it better to create two questions, if you're interested in both?

oh, and of course if optimizing for speed, memory efficiency should also be a factor

Two challenges wouldn't be considered duplicates if they have sufficiently different scoring mechanisms (and size and speed would be sufficiently different). Personally I'd rather see them as separate challenges than all the answers mixed together, so I can compare like for like. I'd still like to read through both sets of answers, just separately.

two different winning criteria in the same challenge doesn't sound very good... :P

Yes, good point.

I think there might've been a few challenges that tried it but weren't well received

CBF to go find an example of such

exactly because two different winning criteria isn't one and only one winning criterion, although they can be if one has a lower priority than the other

Another possibility is a single scoring method that takes into account more than one aspect of the answer (such as points for speed plus points for size, in a well defined formula) but it would be difficult to fine tune this to make it interesting (and even then it may reduce to one measure or the other for some languages)

Yes, that would indeed be difficult to fine tune.

Sometimes a challenge scores on one and restricts on the other. The simplest example is shortest code excluding brute force, by having a limit on run time for a particular input

It actually seems to me that there should be two criteria of size-optimization, which could go into separate challenges. One for using whatever standard libraries are available, and one for using only the basic libraries necessary for I/O and implementing the rest from scratch.

But it could be quite hard to define the latter well enough.

@trichoplax it's usually a set of inputs, otherwise hardcoding might just happen to still be shorter than making your approach faster

You can compare language X with library Y and language X without library Y as if they were two different languages

@EriktheOutgolfer Good point :)

also, I'm a bit against "you can't import from the standard library", because the standard library is just another part of the language, although you may require that there not be any "external" libraries, but that's only imho

@trichoplax "My language is C-with-all-the-imports-already-imported. It's kinda like C"

^ that can only happen with command-line arguments (and I don't think GCC can #include from there... :P)

@Veskah Since you could easily write a compiler for that as a distinct language, it seems fair to allow it

At least, the meta answer on linter-golf seems to support that

Are there any other programming languages where, like unary regex, the maximum value of any integer variable is limited to not exceed the largest number in the input? I'm expecting the answer is no, but I do find it an interesting limitation to operate inside of.

@DJMcMayhem Holy crap, +400 bounty! xD Thank you so much! And If someone else manages to take that for an answer even better than mine, I won't mind!

um, you're being a bit modest here... :P

honestly, a challenge that involves proper divisors is the direct opposite of regex

I also have a regex that calculates an irrational number. And I'm pondering exactly what kind of question I should post so that I can post my answer on it.

And thinking about if it's possible to calculate a transcendental number in regex.

By "calculate an irrational number", I mean it returns floor(input / irrational number greater than 1) as a match.

and... how are you going to represent "irrational number greater than 1" in unary?

That's the challenge of it!

You can't.

then how do you have the regex lol

That's why the task is to divide by it, rather than multiply by it.

but doesn't it have to be given as an input?

or is it hardcoded in the regex?

The irrational number is a constant.

The regex is 849 bytes.

And that's using ECMA + molecular lookahead. I figure that it might roughly double in length when I get around to porting it to pure ECMA.

And that 849 is fully golfed. It started at 1159.

"is fully golfed" yeah, you miiiiight want to retract that decision in the future... :P

Which decision?

Oh! You mean others might be able to golf it even more?

I'd welcome that!

the "is fully golfed" part... you have an 849-byte behemoth, and you may be fully done with it, but that doesn't mean it's ungolfable ;)

Of course.

But by that I basically mean that I went through 39 versions.

And in the latest one, I can stare at it for a couple hours and not come up with any further ideas.

I know the feeling

your brain has been tired of it, but it has happened to me that a fresh brain has golfed the code even more afterwards :P

I'm very happy with the constructs I came up with to make this regex. I hadn't done fully generalized multiplication and division beforehand, which can accept 0.

of course, if there are fresh brains willing to tame that behemoth...

So yeah, do you think posting a question where this can be an answer would be a good way to go about that?

Heck, it'd probably take those fresh brains an afternoon to understand what it's doing. That's a lot of pattern matching

@Deadcode wait so it handles division by zero as well?

No, but it can handle division of zero.

oh, I thought you meant that it had some sort of match fail for that :P

Well it probably would.

But I do love how 0%0=0 is a match in regex.

Precisely because regex is primitive recursive. It instantly drops out of an unbounded loop after a zero-width step.

what is there to take away from nothingness? ;) :P

I'd like to see some step counts on that thing. Like just a 40 byte datetime parser gets like 60 steps

Step counts on the irrational-dividing regex?

Yeah

For what value input?

the counter might get exhausted, mind you...

@Deadcode Not too concerned on that, just a random sample

I'll have to add a feature to print the step count upon finishing. The debug trace is so long I can't just "| tail" it

Heh

input 1000: 8499434 steps with -O2 full optimization enabled, 31874764 steps with -O0

Input 2000: 37726865 steps -O2 / 171399840 steps -O0

Input 4000: 141956012 steps / 859280603 steps

So what's the O()? :)

Input 8000: 580927367 steps / 4615389525 steps

Dang son

Is there any way to put up a bounty for anybody who can outgolf me in any one of a given set of questions?

@EriktheOutgolfer

@Deadcode ^ This is the closest we have, which seems to work pretty well

@trichoplax Thanks :)