[vtk-developers] VTK strings

[David Gobbi]

For the most part, VTK strings are in the local 8-bit encoding, whatever
that happens to be.  On Linux and Mac, the local 8-bit encoding is pretty
much guaranteed to be utf-8.  On Windows, if you're in North America or
western Europe, its latin1 or more precisely Windows-1252.

The reason this is so is that the IO classes (readers, writers, etc) simply
use 8-bit strings filenames etc. when calling system IO functions.  VTK
uses ifstream(const char *fname, ...) and let's the system decide how
"fname" is encoded.  But this is not consistent across all of the readers,
since some readers use third-party libraries to handle the IO and then
you're at the mercy of whatever encoding that third-party library uses.

On the display side of things (e.g. when using the VTK text mapper classes,
I believe that VTK actually does use utf-8, but I haven't experimented to
be sure that all the VTK display classes work the same.

In other words, strings are a bit of a mess in VTK unless you're willing to
be satisfied with ASCII.

The vtkUnicodeString is UCS-4 (32-bit code points).

 - David


On Thu, May 10, 2018 at 5:35 PM, Todd via vtk-developers <
vtk-developers at vtk.org> wrote:

> Can someone please tell me the default/expected encoding for a std::string
> in VTK. I'm assuming it is UTF8. Therefore I expect vtkUnicodeString (a
> terrible name) is encoded as UTF16. Is that correct?
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://vtk.org/pipermail/vtk-developers/attachments/20180510/dbb56b69/attachment-0001.html>