12 Jun
2017
12 Jun
'17
2:17 p.m.
On 12 June 2017 at 15:11, Peter Dimov via Boost
Artyom Beilis wrote:
Converting invalid UTF-16 to WTF-8 and other way around is not obvious
behavior and has potential of security risk especially for users that are not aware of such an issue. So invalid UTF-8/16 sequences are rejected by design.
Implying that there are Windows file names that can't be handled by the library?
Question: "Shouldn't the passing of invalid UTF-8/16 sequences be defined as UB?" How important is this point? I'm oblivious to this problem, and it sounds like I would like to keep it that way. degski -- "*Ihre sogenannte Religion wirkt bloß wie ein Opiat reizend, betäubend, Schmerzen aus Schwäche stillend.*" - Novalis 1798