Thread by @FakeUnicode

The savvy Twitter user knows to always try to use #PascalCase (upper camel case) for hashtags, since hashtags are (ideally) case-insensitive: #MayThe4th

The case insensitivity is such that these should all be aliased:

#SSSs #SSsS #SSss #SSẞ #SSß #SsSS #SsSs #SssS #Ssss #Ssẞ #Ssß #sSSS #sSSs #sSsS #sSss #sSẞ #sSß #ssSS #ssSs #sssS #ssss #ssẞ #ssß #ẞSS #ẞSs #ẞsS #ẞss #ẞẞ #ẞß #ßSS #ßSs #ßsS #ßss #ßẞ #ßß

Hashtag length is 100 characters:

#ssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss

Can't do (on twiweb anyway):

# + sssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss

But what if:

#ẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞẞ

That decomposes/aliases/casefolds to two hundred "s".

Hashtags also alias with different "fonts" (mathematical alphanumerics, fullwidth): #s𝐬𝑠𝒔 #𝓈𝓼𝔰𝖘 #𝕤𝗌𝘀𝘴 #𝙨𝚜ｓｓ

And with many diacriticals that have decomposition: #ŚśŜŝ #ŞşŠš #ṡṢṣṨ

But does it alias with hashflags?

#stayhome
#𝙨tayHome
#ṢtayHome

Looks like hashflag (fake emoji) insertion only does case folding, not NFD.

Also, astral characters in hashtags are still broken for index counting on Twitter Web. Hashtag insertion is done as Position + Length and it's counting length in UCS-2 Units. So you get �s.

Modest (serious) proposal for Twitter: Ignore underscores in hashtag search.

Like how GMail ignores fullstops in emails (foobar@ foob.ar@ f.o.o.b.a.r@ identical).

So that #ssss and #SSSS and #s_s_s_s are all identical.

Then people have a choice of camel casing or underscoring.