Strupremum

With regards to word validity, I think it would be best to have a built-in dictionary and provide the ability for players to vote on the validity of a word.

DJMcMayhem

^

flawr

5:13 PM

@DJMcMayhem go ahead

Mithrandir

a wordlist set will never be comprehensive

Pavel

I could go

flawr

@DJMcMayhem never ending regex? :D

Erik the Outgolfer

maybe allow Google for checking validity?

DJMcMayhem

Nope, just ~~carrot~~ caret

Erik the Outgolfer

5:13 PM

although that might lead to... accidents

El'endia Starman

I've got a rule in mind.

Pavel

Rule: .*([abc])([def]).*\1\2.*

ninja

Oh, sorry, .*, not ^

reset timer? :P

5:16 PM

Oh, that's very different

Erik the Outgolfer

phew, waaaay easier

DJMcMayhem

We should also clarify whether anchors are implicit

Erik the Outgolfer

they're not; the end

Mithrandir

What do \1 and \2 mean?

El'endia Starman

From what I've seen, it looks like every regex has been anchored.

Pavel

5:17 PM

@Mithrandir Backreferences

Erik the Outgolfer

backreferences

flawr

13 mins ago, by quartata

effervescence (13)

Mithrandir

....which means?

Pavel

\1 matches the first thing in (), \2 the second

Erik the Outgolfer

lol ce...ce

DJMcMayhem

5:17 PM

@flawr I don't think that fits

Pavel

So ([abc]) and \1 must be the same letter

Erik the Outgolfer

I think it does

> effervescence

DJMcMayhem

Oh, the ce part, not the fe part. Yeah, that fits

Erik the Outgolfer

of course, it must be spelled right

@flawr ^ (13)

flawr

sorry

typing is hard

Pavel

5:19 PM

My word is 10

El'endia Starman

@Mithrandir In a regex, () denotes a capture group. Oftentimes they are just used to group things that should be repeated but they can also be used for backreferences, as mentioned. \1 refers to the first such capture group, \2 the second, and so forth.

J. Sallé

The word needs to have one of [abc] followed by one of [def], twice. Is that right?

El'endia Starman

Right.

J. Sallé

Okay

DJMcMayhem

@J.Sallé But they must be the same each time

J. Sallé

5:20 PM

Yeah I got that

DJMcMayhem

Gosh, I can't even think of any short ones

J. Sallé

I had a long one with 2 ab >.>

Pavel

I thought of a whole bunch that all follow a pattern just now

flawr

worcestershire-sauce?

would that count?

DJMcMayhem

I think so

El'endia Starman

5:23 PM

Abusability (11) <- does that count? Or maybe "abuse-ability"?

Erik the Outgolfer

um

no

ab isn't a valid pair

El'endia Starman

Wow, how do I keep making these simple mistakes.

J. Sallé

@flawr That's 2 words, no? Not hyphenated

flawr

idk that is why I'm asking

Pavel

Hyphens are allowed either way

Erik the Outgolfer

5:24 PM

. isn't only letters...

you want [a-z] otherwise

DJMcMayhem

@EriktheOutgolfer By that logic, sentences work because they match space and punctuation...

flawr

yeah

J. Sallé

Yeah but what I mean is the name of the sauce is "Worcestershire sauce", without a hyphen, isn't it?

I've never seen the name hyphenated

Erik the Outgolfer

> players post the longest word they can think of that matches

hyphenated words are still words, sentences aren't

flawr

In german it is without hyphen or space :)

Pavel

5:25 PM

I think it's time

J. Sallé

If we were playing in German I'd probably be doing better >.>

I had Incandescence btw, but that's also 13.

flawr

@Pavel what word did you have in mind?

@J.Sallé oh you speak german?

Pavel

Here's what I came up with:
Bedchamber
Radiobroadcasting
translucence

Erik the Outgolfer

you sure you didn't brute-force them?

the change in case looks a bit suspicious

J. Sallé

@flawr some German, yes. I studied Hochdeutsch up to B1 before I started working.

Pavel

5:27 PM

Well I picked out bedchamber and radiobroadcasting as random words

And then found a pattern to fit them

flawr

@J.Sallé oh cool!

Pavel

I came across translucence later

J. Sallé

I still plan on getting to C2 eventually, just need the time to do it :p

flawr

Rule: ..r..r.*

Pavel

Anchored?

Mithrandir

5:29 PM

@El'endiaStarman Alright. I'm relatively proficient in regex (I did maintain the IPS Comment Bot regexes almost single-handedly for a long time), I've just never encountered backreferences before. Thanks for the explanation.

Erik the Outgolfer

berserk (7)

berserker (9) (valid?)

DJMcMayhem

Or berserking...?

flawr

~~ninja~~

DJMcMayhem

@flawr Is that anchored?

Erik the Outgolfer

I don't think "berserking" is valid... is it?

DJMcMayhem

5:30 PM

Berserk is a verb, right?

flawr

@DJMcMayhem uhm what does that mean?

Mithrandir

Never seen it conjugated like that before

El'endia Starman

I think beserking is valid, yes.

Erik the Outgolfer

"beserking" surely not... ;-)

DJMcMayhem

@flawr Does it have to start with ..r or can that be in the middle?

Strupremum (10)

:P

flawr

5:31 PM

@DJMcMayhem ah you mean the ^$

@DJMcMayhem guess where I've got the idea:)

but I doubt it is an official word

(yet)

quartata

Dec 14 at 3:15, by quartata

if you manage to win with "strupremum" anywhere you are immediately crowned champion and showered with likes and retweets

Erik the Outgolfer

yeah

quartata

theres an exception to the dictionary word rule

DJMcMayhem

Harboring (9)

flawr

@DJMcMayhem I think the whole word should be matched anyway, right?

J. Sallé

5:32 PM

Murderous (9)

Erik the Outgolfer

lol "berserking" is our best bet?

DJMcMayhem

@flawr Technically, that regex would match anything that ends with ..r..r.*

Barbarians (10)

Erik the Outgolfer

barbering (9)

flawr

time's over

@DJMcMayhem really?

I thought it would only match this part of the word (and not the whole word)

Erik the Outgolfer

yeah, a match doesn't need to be "whole"

quartata

5:34 PM

no

DJMcMayhem

It would only match part of the word, but it still "matches the string" if that makes sense

quartata

it needs to be whole

DJMcMayhem

Did I win?

flawr

@DJMcMayhem yeah that is what I thought, but I expected that here we want the whole word to match

DJMcMayhem

@flawr What was your word?

quartata

5:35 PM

we're looking for the largest string in the language defined by the regex

DJMcMayhem

@flawr OK, that's fair

Erik the Outgolfer

@DJMcMayhem well... didn't you suggest that "berserking" thing? :P

like, a second before my mind went there?

flawr

@DJMcMayhem strupremum, as I mentioned before :)

@DJMcMayhem yes

DJMcMayhem

@EriktheOutgolfer I was suggesting it as in improvement to your word, and then I got a 10 afterwards

Erik the Outgolfer

ah, alright

rule change request: suggestion time counts, not validation time

DJMcMayhem

5:36 PM

Although you technically never accepted/posetd my improvement :P

Erik the Outgolfer

although I might have to study now

flawr

@EriktheOutgolfer I'm sure you don't

Erik the Outgolfer

hm... algebra exam tomorrow... maybe not :P

DJMcMayhem

How can I say in regex "A letter followed by a different letter"?

Like (.)(\1)\@!(.)?

flawr

(something)[^\1] maybe?

Cows quack

5:38 PM

negative lookahead

Erik the Outgolfer

lookahead is prohibited

El'endia Starman

Oh, but we decided early on not to have lookaheads or lookbehinds.

DJMcMayhem

@flawr Unfortunately that doesn't work

El'endia Starman

They make things way more complex.

Cows quack

otherwise I don't know how you would do it other than a[b-z]|b[ac-z]|c[abd-z]|...

El'endia Starman

5:40 PM

That would work, yeah.

Erik the Outgolfer

yeah, I don't think it's possible without lookahead... but that works :P

DJMcMayhem

Actually (at least in vim): (.)(\1)\@!(.) does work afterall

El'endia Starman

Is \@! a vim thing? I haven't seen it before.

DJMcMayhem

Yep

Cows quack

'Like "(?!pattern)" in Perl.' (from vim docs)

Erik the Outgolfer

5:42 PM

hm...

Pavel

No lookaheads or lookbehinds allowed

Erik the Outgolfer

no vim-regex please...

Pavel

Since (?!...) isn't allowed neither is \@!

Cows quack

3 mins ago, by Cows quack

otherwise I don't know how you would do it other than a[b-z]|b[ac-z]|c[abd-z]|...

DJMcMayhem

> no lookaheads, but capture groups and backreferences are allowed

Cows quack

5:43 PM

so all dj has to do is to post a 100+ byte regex now

DJMcMayhem

It doesn't say anything about lookbehinds

quartata

no lookbehinds

Erik the Outgolfer

>_>

quartata

the only reason why capture groups are allowed is i was buillied into it

Pavel

lord quart has spoken

Erik the Outgolfer

5:44 PM

lookbehinds are just lookaheads with the eyes turned the other way

DJMcMayhem

I've got a really simple pattern, why does it have to be so hard to describe in regex?

J. Sallé

Regex in a nutshell

Skidsdev

I think capture groups are fine, but lookaheads/lookbehinds are definitely a bit too complex for brain-regex

Erik the Outgolfer

regex should really have a, you know, not lookahead NOT operator!

Pavel

There does need to be a not

quartata

5:45 PM

Boolean grammar

Boolean grammars, introduced by Okhotin, are a class of formal grammars studied in formal language theory. They extend the basic type of grammars, the context-free grammars, with conjunction and negation operations. Besides these explicit operations, Boolean grammars allow implicit disjunction represented by multiple rules for a single nonterminal symbol, which is the only logical connective expressible in context-free grammars. Conjunction and negation can be used, in particular, to specify intersection and complement of languages. An intermediate class of grammars known as conjunctive grammars...

you wanna?

Erik the Outgolfer

similarly an AND

DJMcMayhem

And [^\1] doesn't count?

Erik the Outgolfer

but it doesn't work

El'endia Starman

@DJMcMayhem Same reason that parsing HTML with regex opens a portal for eldritch abominations. :P That is, the language(s) we use is more complex than that which is parsable by typical regexes.

Erik the Outgolfer

[^\1] literally means "not chr(1)"

Skidsdev

5:46 PM

He Comes

Erik the Outgolfer

who comes?

Pavel

No recursive or balancing groups either

So no palindromes

Erik the Outgolfer

palindromes are recursive, yeah

J. Sallé

@EriktheOutgolfer Helium

DJMcMayhem

Rule: .*(..)\1.*

El'endia Starman

5:48 PM

No arbitrary-length palindromes, that is. You can still specify that a finite subsegment of a word be palindromic.

bananas (7)

Ananas (6)

LOL

goddamnit erik

XD

5:49 PM

do song lyrics count too? :D

Erik the Outgolfer

not if they have spaces or newlines in them...

:P

Skidsdev

Does a welsh word count?

flawr

@Skidsdev I doubt as already german words do not

Erik the Outgolfer

that would be unfair

El'endia Starman

For the sake of our sanity...no. :P

J. Sallé

5:50 PM

@Skidsdev fair reminder that Ll is a single letter

Erik the Outgolfer

^ Spanish too

alright I hope "bananas" isn't the best we can come up with

Skidsdev

@J.Sallé yes, but the word I'm sure we're all thinking of in welsh ends in gogogoch, which satisfies the rule

J. Sallé

Pretty sure it's a digraph in spanish but I might be wrong

@Skidsdev it's that town's name isn't it?

Skidsdev

yes

Pavel

murderers (9)

Erik the Outgolfer

5:52 PM

balalaikas (10)

Cows quack

are chemical element names valid?

Erik the Outgolfer

I think so

Cows quack

ununpentium 11

Erik the Outgolfer

but not "unun" :P

it's not called "ununpentium" anymore...

J. Sallé

Balalaika is the name of a (very, very bad) vodka brand in Brazil rofl

Erik the Outgolfer

5:53 PM

so... do we count archaic words?

El'endia Starman

IIRC the rule on names (and proper nouns in general) is basically that they have to be really widely used. So Mississippi is fine but specific cities or towns probably aren't.

J. Sallé

I think if it's in a (recent) dictionary, why not?

Erik the Outgolfer

I wish Mississippi was really valid...

El'endia Starman

@EriktheOutgolfer It was at one time. And not that long ago either.

Erik the Outgolfer

ik

but do we want to count it?

Cows quack

5:54 PM

we need to have a specific dictionary