The Nineteenth Byte

12:36 AM

@ChartZBelatedly I think this belongs to the "the question doesn't actually explicitly say to not do this, but it's obvious that this is against the spirit of the challenge, and a 3k rep user should know better than to exploit a new asker's inexperience" category.

Maybe I should've just deleted the post alongside the user. It's a pretty simple kolmo tho, so I didn't think anything could go wrong with just leaving it here since people had already posted upvoted answers.

Deadcode

@Neil 39 bytes, not 46

How did you get such an arbitrary number 46

Neil

@Deadcode Try it online! but replacing the r modifier with rewriting the regex in left to right order

(which seems to work fine, contrary to your reply to my comment on your answer)

Deadcode

@Neil That's not a pure regex if it uses r

Neil

well, r just makes it match right to left

if I rewrote it into a left to right regex it wouldn't need the r

but it would be a byte longer because it would swap lookaheads and lookbehinds

Deadcode

Yeah, that's my point. You need to do that.

But anyway, I got 39 without any .NET specific features.

Neil

12:49 AM

well, removing the r would save two bytes, so I'd end up with 46

Deadcode

Bytes? How do you count flags in byte counts?

If you're counting it in Retina bytes, then it's definitely not a pure regex

Neil

that link is to a 47-byte Retina program

Deadcode

Also BTW I found it very strange that you posted an answer with just a "Generic" regex. You almost never do that. It was very unusual to see a post of yours that isn't .NET-specific

Neil

if I were to preform the rewrite into a pure regex then it would have been 46 bytes, but I was too lazy because it's too tedious

and in fact I can copy your ^()? trick to get it down to 40 bytes anyway

Deadcode

no, you can use my regex unmodified to make a 39 byte Retina program

but then it's unary

and you don't do that, so I guess it'd be 45 bytes

so I guess if you have a 40 byte up your sleeve it's probably using Retina-specific things

Neil

12:53 AM

ok, let's start again

Deadcode

But you would actually copy my regex without even upvoting it? That's cold

I upvoted yours

New Main Posts

1

Q: How often will this test fail?

Preamble There was a unit test in our codebase which was shuffling a string of [A-Z]+[A-Z] and then using the first twenty characters of that shuffled string. It failed quite a while after being written due to no repeated character being present in those twenty and we wondered how often such a te...

Neil

I posted a 50 byte regex in my comment

although what I had actually tested with was the 47 byte retina program that I linked to above

Deadcode

but you said 46 bytes

@Neil Also you always include decimal-to-unary conversion in your Retina programs, so it's unfair not to include it in this one

your TIO link doesn't include it

it puts it in the header

Neil

right, but I was imagining that I could turn it into pure regex, which would avoid the r modifier

although now that I double-check it would have been 47 bytes anyway

then using your ^()? trick reduces that to 40, and then I've since seen where you get the last byte from ((x|y)1) can be (x1|y1)

Deadcode

12:57 AM

40 with unary conversion or without it?

because if it's without it, why are you saying 40 instead of 39?

Neil

the retina program was 47 bytes because I was too lazy to do it as pure regex, it was just as a proof of concept that a conditional would be shorter

OldSandboxPosts

Sandbox posts last active a week ago: Limit of lists

Deadcode

@Neil the elephant in the room is you're talking about copying my trick and you haven't even upvoted the post. What's up with that?

Neil

ironically that's not the trick I'm actually going to copy for my post

@Deadcode that was just me commenting that my approach to an optional capture was suboptimal anyway which makes the whole question of byte count moot

so now the only differences is that our tests are in the opposite order and you've grouped yours and wrapped the entire group in ^$ while I just kept each alternation with its own ^$ (and I'm using 1s instead of xs)

btw, nice ES2018 workaround for lack of conditionals or failed backreferences

(gap in between my last two messages was how long it took me to figure out how it works)

Deadcode

1:31 AM

@Neil Thanks.

@Neil Oh, I'm an idiot. I tried your regex in my test harness without changing the 1s to xs or changing the xs to 1s in the harness. My apologies! Your 50 works without the r flag.

Sorry for the confusion

New Main Posts

0

Q: Ball game King of the hill

Ball game is a game I invented where two players start with 100 energy each. Every turn, each player independently chooses an amount of energy to spend. Whoever spends more energy brings the ball 1 unit closer to their side. If the ball is on someone's side, 4 units away from the center, the pers...

king-of-the-hill javascript

Deadcode

I've mostly left it unsaid, but: I like that your convention of using 1 makes it actually literally bijective unary using the standard numeral glyph(s). But the practical downsides are that it's less human-readable (they blend in with backrefs), and you can't put them directly after a backref...

which in the best of circumstances merely forces you to reorder things at no byte cost (but sometimes resulting in worse speed) but occasionally might actually prevent a certain approach from being usable, or cost actual bytes.

Neil

yeah, that's probably why Retina 1 switched to _s

Deadcode

@Neil strangely your 50 byte works with and without RegexOptions.RightToLeft

Neil

1:47 AM

right, the 47-byte program is simply with the lookbehind removed and forcing the option

on an unrelated note, is 30 the smallest integer with representations in three different bases that contain exactly two 1s and no other non-zero digits?

Deadcode

Wait, the r flag should never make any difference in single-number unary, right?

Does it not essentially evaluate the regex as if the string had been reversed and the regex were done forward?

@Neil Ah, thanks for that link. That's rather disillusioning... all this time I had believed the Mozilla team implemented their own lookbehind from scratch. But they outright admitted porting it from V8

Wasif

2:14 AM

Good morning!

Sandbox

Transcript for