Tavern on the Meta

« first day (1946 days earlier) ← previous day next day → last day (3058 days later) »

Doorknob

10:00 PM

A rite of passage for any new site.

user259867

I forgot one can downvote in private beta with association bonus...

user259867

And even vote to close. But not edit.

Braiam

I'm looking for the status-declined of the FR on SF...

user259867

ɆnħȺnȼɇđ Mᵾsȼłɇ Ƀᵾɨłđɨnǥ

user259867

It's easy to find a tool for obfuscating text like so, still don't see a good one for the reverse process.

Braiam

10:10 PM

@tchrist is there a way to programmatically remove zalgo of text?

I presume there is, but I'm more interested on the gottchas

tchrist

@Braiam Trivially.

Convert to NFD, delete all \p{Mark} code ponts.

You’ll ruin Spanish though.

user259867

NFD

NFD (meaning Noise for Destruction) are a London-based hybrid metal band formed by Peter "Bob" White and Simon Rippin (both formerly of Sensorium) and Tony Pettitt (formerly of Fields of the Nephilim). The band have released three albums and three EPs. == History == The band was formed after Pettitt had left the aborted Fields of the Nephilim reformation in 2002. Together with drummer Simon Rippin (ex the Nefilim), guitarists Chris Milden and Stephen Carey and vocalist Peter White (ex Sensorium) they signed to Jungle Records in 2003. They made their debut with the three song Break The Silence EP...

Braiam

brb, I need my guitar

user259867

But it's not necessarily combining characters: they also use things like 𝔼𝕟𝕙𝕒𝕟𝕔𝕖𝕕 ๓ยรςɭє Ｂｕｉｌｄｉｎｇ

tchrist

The best way to do this is to compare at Unicode Collation Strength 1.

That way things test equal.

For example, at UCA 1, all these are equal to the UCA1("A"):

U+00041 ‭ A  LATIN CAPITAL LETTER A
U+00061 ‭ a  LATIN SMALL LETTER A
U+000AA ‭ ª  FEMININE ORDINAL INDICATOR
U+000C0 ‭ À  LATIN CAPITAL LETTER A WITH GRAVE
U+000C1 ‭ Á  LATIN CAPITAL LETTER A WITH ACUTE
U+000C2 ‭ Â  LATIN CAPITAL LETTER A WITH CIRCUMFLEX
U+000C3 ‭ Ã  LATIN CAPITAL LETTER A WITH TILDE
U+000C4 ‭ Ä  LATIN CAPITAL LETTER A WITH DIAERESIS
U+000C5 ‭ Å  LATIN CAPITAL LETTER A WITH RING ABOVE
U+000E0 ‭ à  LATIN SMALL LETTER A WITH GRAVE
U+000E1 ‭ á  LATIN SMALL LETTER A WITH ACUTE
U+000E2 ‭ â  LATIN SMALL LETTER A WITH CIRCUMFLEX

Notice the stuff at the bottom.

I know how to do this with Perl because of the Unicode::Collation module. But you have to compare strings not patterns.

So you can do your string search at UCA1, for example.

Then they would not escape.

Braiam

10:26 PM

latest rave on es.SO "how to validate emails with regexes?"

man... you shouldn't use regex!

tchrist

No?

Why not?

I have a full grammar for that.

Braiam

because, you should just send the email and wait for the bounce

if the mail is valid, it will be fine, if not, who cares?

tchrist

president@whitehouse.gov is valid.

And useless.

Braiam

exactly

tchrist

The UCA at primary strength compares only "basic" letters, ignoring nonletters. It ignores case and diacritics as well.

So the math letters they are using to avoid Smokey test as the same as regular letters at UCA1.

user259867

10:30 PM

Or, instead of decoding the obfuscated text, we could take the fact of obfuscation as a spam signal...

tchrist

Oh yes, surely: "ransom note detected".

user259867

Ordinary users 𝕕𝕠𝕟'𝕥 𝕡𝕠𝕤𝕥 𝕨𝕚𝕥𝕙 𝕥𝕚𝕥𝕝𝕖𝕤 𝕝𝕚𝕜𝕖 𝕥𝕙𝕚𝕤

user259867

Except maybe on Math, if they copy-pasted the "blackboard bold" text from elsewhere.

tchrist

ducks

Just make sure my top SO post doesn't summon the gendarmes.

user259867

There's 50 rep limit for Smokey reports.

user259867

10:33 PM

Except it doesn't apply when the title is received through websockets, because reputation isn't known then. Hmm.

user259867

@tchrist But your post is an answer, and the title doesn't have any of that stuff.

tchrist

Oh right, I was thinking of something else.

I have emoji in some MSE questions, but not much. Nothing like the headers in the SO answer.

I was thinking of h1 h2 etc as title, wrongly.

Ok yeah, do that: count math letters as a strong signal. Maybe do proportion or something.

Much easier than switching over to UCA1 string-search comparisons when most everything now is regexes.

Braiam

I presumed smokey was plugable

tchrist

Hm, I wonder whether this was brought up on PT.SO early on:

Sería interesante averiguar si en los sitios en portugués, japonés y ruso han tenido la misma discusión, y en tal caso cuáles han sido las conclusiones. No para copiarlas ciegamente, pero al menos para tener algo de perspectiva. — Konamiman 51 mins ago

I'll go look on meta.

Q: Tradução de perguntas e FAQs

Qual política deveremos adotar quanto a tradução de perguntas "famosas" ou mesmo pertencentes aos FAQs do SO.com original? Como por exemplo: http://stackoverflow.com/questions/236129/how-to-split-a-string-in-c entre outras do tag [c++-faq]? Devemos pro ativamente começar a traduzir? Retorne p...

Which Gabe answered.

SmokeDetector

11:02 PM

[ SmokeDetector ] All-caps title: [HARD PROBLEM?] by user29557 on crypto.stackexchange.com (@Unihedron)

cybermonkey

just raised my first 'proper' flag on AU

user259867

11:21 PM

Reopening this. It's nominally on-topic, and although I'd advised the room regulars to let it drop, they appear to be all about continuing the argument as long as possible - therefore, I'm using their responses here as input in a proposal I'm drafting for new chat guidelines. As the old saying goes: when life gives you lemons, rub them all over and enjoy the sensation. — Shog9 ♦ 2 hours ago

tchrist

What would SO be like if Programmers didn't exist?

user259867

Same thing, I guess. Programmers site is too small to matter.

user259867

I mean, if Programmers disappeared today. If NPR was never created, that would be somehow different, but I wasn't around then.

UNICYCLEZ R DE BEST IN DE HERE

I feel like slapping someone

user259867

11:54 PM

user259867

Interesting: the API returns the 0 reputation "changes" associated with downvoting questions.

user259867

For historical reasons, I suppose: they were not always free.

user259867

I was testing to see if full-reputation-history is the method that leaks the posts on which votes by a deleted user were vacated. But it's been a long time since I lost any points that way.

tchrist

@Braiam Acabo de enviarte un mensaje mágico. :)

user259867

Could someone please upvote one of my posts and then delete your account? For science.

user259867

11:58 PM