Juski, aka yuscii, was the codepage used on Vaha, pdp, Partner and then many PCs (including all of those sold by DBA during my time). Analogue to similar german or other "codepages", where the non-english characters were crammed in the slots taken by brackets, braces and other punctuation. The string "{|~}`[\^]@" was actually to mean "šđč枊ĐČĆŽ". In the days before soft fonts (EGA, VGA and later), the letter shapes were burned into ROMs of the video cards and you pretty much had to have your own, so the 100 Hercules cards we got from Acrotech actually had both sets - american latin and yuscii latinic, and there was a switch on the card that flipped between the two. So „ŠANTIĆ“ would become „“[ANTI]“ and vice versa.
The use of backslash (which was m$ caprice, Gates wanted to be different from the Unix world) for directories and for Đ in yuscii had often created words which seemed impossible to pronounce. C:\Fina would be C:ĐFina, but... we actually pronounced đf and many other combinations without a problem.
When parting with DBA we switched to CP 852 (in Avai and Szoftex), then under windowses to 1250 and then under linux I switched to UTF-8. However, we had to keep yuscii in several places where they still used hercules cards. Many newspapers stick to it even nowadays, because they use Quark or some such app, which is still not Unicode. Additionally, most of the state institutions have some legacy apps, e.g. examples from cadaster that Marinko sent me were even yuscir - which is yuscii cyrillic, with everything the same except q stands for lj, w for nj and x for dž. This was a source of many internal jokes - the divx format accidentally printed in cyrillic as divdž, windows would be njindonjs, and njalter actually used that intentionally in his username.
There was trouble in collation as well, which was properly solved for larger languages only at different times. The replaced characters didn't get into their proper places, so Ž, the last one in serbian latinic, was actually first, and the sequence c-č-ć-d-đ-dž was all out of whack; furthermore, the two character sequences like lj, nj, dž were treated as two separate characters, so words beginning with lj would not come after the last one beginning with l, but rather after li and before lk (ok, nothing begins with lk, so before lo). Even nowadays (2023) it's not completely solved - while in the databases there is a proper serbian collation, not so in programming languages, so even Python uses russian collation for serbian (not quite language aware) so š is not the last cyrillic character, as đ, j, lj, nj, ć and dž come after it.
I wonder how did Hungarians fare at the time of huscii - we have these five (plus uppercase); they have nine and they're all vowels. Also, what was going on with dutch diphtongs, did sj come after sz and before t, or did it go between si and sk... which brings a worse problem with hungarian, where dzs could be d+zs (d+zh, when agglutinating) or it could be dzs (a jay, i.e. dž). Trouble all over, and yet the creators of everything new in computing and specially anything that outputs characters (on screen, plastic tape, paper, projector, decals, cutout letters with magnets, alphabet soup etc etc) would invariably do it only for all haircut set first, then for the rest of the world later, if ever.
11-IV-2023 - 24-XI-2023