The following documentation comment has been logged on the website: Page: https://www.postgresql.org/docs/11/unaccent.html Description:
Seems a bug, because the `select unaccent('Iglésias')` result in accented "iglésias" again... It is correct because length('Iglésias') is 9 instead 8. The problem is not rare as you can check by pageviews of https://stackoverflow.com/questions/24863716 The solution is to feed database with good UTF8 (NFC characteres). **SUGGESTION**: add a notice for reders, about the aparent bug with non-NFC input, showing examples and clues about solutions. REF: https://en.wikipedia.org/wiki/Unicode_equivalence#Example