search for: surr_good

...rrect.? AFAICT, the current handling attempts to decode a surrogate pair by combining the high surrogate with the same high surrogate, instead of the high surrogate with the subsequent character that hopefully is the low surrogate[7]. Here is some code that could be added to regression tests: ??? surr_good <- '\ud840\udc00'??????????? # auto-converts to normal ??? surr_bad <- paste0('\ud840', '\udc00') # surrogates remain ??? good <- c('hello', 'world', surr_good, '\ud7ff', '\ue000', '\U0010ffff') ??? bad <- c(surr_bad, '...