[Octave-bug-tracker] [bug #52905] Can't manage paths that contain extended unicode characters

[anonymous]

Follow-up Comment #3, bug #52905 (project octave):

Well, it does not look like an API misuse in the GUI binary.
Windows GUI shell opens "unicoded" filenames without questions.

It seems to me, the problem is in interpreter. It looks like standard
path/name handling in such functions as cd(), or readcvs/dlm do not  handle,
or do not allow handling strings (probably corrupting them) as Unicode when
passing to Windows system. 

You see, "ＴＥＳＴテスト①.csv" looks like
exact 8-bit conversion of "ＴＥＳＴテスト①.csv", so there is no data
corruption, only data-type is undetermined in Ocave, since there is no
UTF-char type.
However, (AFAIK) WinAPI functions expect the data in specific format for
further handling. And I foresee three possibilities:
1)Octave cripples the string with precaution procedures when passing it to
WinAPI. I.e. like truncating or extending the string basing on difference in
byte/character count. Or replacing "unknown" bytes with "?"...
2)Octave misuses WinAPI, calling obsolete functions(like winapi OpenFile,
fopen instead of _wfopen)


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?52905>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/