https://bugs.kde.org/show_bug.cgi?id=520509

--- Comment #7 from Daniil B <[email protected]> ---
Hey again!
I've left a comment in the go-keyring issue as well:
https://github.com/zalando/go-keyring/issues/143#issuecomment-4535934154

Again, sorry for the initial panic — I wasn't sure where to report this. After
investigating more, I now believe the primary bug is in go-keyring: it
hardcodes `content_type="text/plain; charset=utf8"` for all data, including
binary. That's simply incorrect on their side

That said, I think `ksecretd` could still be improved a little bit here —
silently corrupting data is the worst possible outcome, as you have mentioned
above. If `content_type=text/plain; charset=utf8` is declared but the data
contains invalid UTF-8 sequences, it would be much better to reject the
operation with an error than to silently replace bytes with U+FFFD

That way the client gets a clear signal that something is wrong, instead of
storing garbage and discovering it later. What do you think about this? Maybe I
could try to fix it too, I wouldn't mind being in KDE contributors :D

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to