Comment 32 for bug 678808

Logged In: YES
user_id=1511901
Originator: NO

Hi Max

Although the best (and I think finally inevitable) solution is to make hugin use only Unicode strings, that will be hard, as Pablo has pointed out. There is still a lot of code that assumes ANSI strings. Newer code uses UTF-8 Unicode, which can "pass" for ANSI so long as one only uses chars in the current codepage; but in cases such as this one the masquerade breaks down.

Finding and fixing all the ANSI code would be a long, mainly thankless task needing very good debugging tools and plenty of motivation. It may be, as Pablo suggests, that on Linux we will eventually be able to build old ANSI code for UTF-8, without actually changing it; but I don't think that will ever be possible on Windows.

So actually the only practical solution for the present bug is to work around it by using ANSI filenames. However, that need not mean changing your Cyrillic names, because every NTFS file does have an ANSI name: the hidden DOS (8.3) name, which you can see with the command "dir /x". That is supposed to be a valid alternative name for the file in all situations, so it might work here.

Regards, Tom