Version: 6.3 Unicode
Happens since: since I'm using this feature, at least 6 months
I've got autocompletion feature turned on, minimal length in my case is 4 characters. I'm using Polish language, which has characters (diacritics) like ą, ę, ż etc. The bug is independent on coding I choose.
Let's take HTML autocompletion as an example. When I write "ali" it won't show code completion (3 chars), after next char ("g") the word's length is 4 characters and it shows list of words, in this case "align". So it's expected behavior.
The bug occurs when I'm trying to write word with Polish diacritics, eg. word "mówi". As there's no HTML keyword starting with "mówi", it shouldn't show anything, but it shows list with "width" selected. It's because it takes length of whole word (4 characters), but completion list shows words starting with characters after "ó", so "wi" - only two.
Another case is with word "przecież" - it shouldn't show anything as well, but I see list with first element selected. Word is 8 char long, it takes chars after "ż" (empty string), so it gives first element. When I'll write przecieżal it will select "align" - 10 chars>code completion is on>takes chars after last whitelisted (?) char (it's "al") and it shows code completion with word that's not contained in list.
mów - nothing
mówi - selected "width"
mówić - disappears because there's no "wić" prefix
mówićx - selected "xml"
It's weird as "przecież" is showing list with chars after "ż" (empty), "mówićx" after "ć", but "mówić" after "ó" - "wić".
It's not critical, but it annoys me as I see code completion list way too much - it don't obey minimum chars count limit. It occurs not only with Polish chars, but with many other as well (like ü, °), while it seems to work fine with some like !, $, @.
Thanks in advance.