Here the pronounced patch:
http://hg.savannah.gnu.org/hgweb/octave/rev/7d6f38541902
SYSTEM encoding is default on windows, UTF-8 otherwise.

Yes, I see the same behavior unter Ubuntu Linux (14.04):

• When I open an ISO8859-1 file with umlauts, those characters are just omitted with Octave 4.0.1+ (default encoding setting = "system default") currently.
• When I open an ISO8859-1 file with umlauts, I get funny diamond shaped question marks instead of the umlauts with Octave 4.0.1+ (explicit encoding setting to "UTF-8") currently.
• Wen I open the samme ISO8859-1file with umlauts in Octave 4.0.0 release, then I also get funny diamond shaped question marks instead of the umlauts.

note: I just now realise, have I misunderstood you? Do you maybe suggest to set the default encoding ONLY for Linux to UTF-8? In this case I would fully agree :) It's better so see funny diamond shaped question marks (then at least you know that some encoding error just destroyed your umlauts), instead of silently dropping several text characters.

...The following text was written when I still thought that you wanted to change the default encoding to UTF-8 BOTH for Linux and Windows:...

Nevertheless I DO like the current behavior a lot - not the funny (but exotic) behavior described above, but the fact that:

• Under Linux OS I can open any m file that was created (under Linux) with Octave 3.8.2 release and Octave 4.0.0 release, and I do see all umlauts displayed properly.
• Under Windows OS I can open any mm file that was created (under Windows) with Matlab, with Octave 3.8.2 [with Octave 4.0.0 I am not sure at the moment, but I could check], and I do see all umlauts displayed properly.

I think the use case that someone save an m file under one operating system, and then opens the saved m file under a different operating system is quite rare and exotic. Whereas I think that the use case that someone under Windows wants to open an m file that was created under Matlab happens (hopefully!) quite often.

This is why I do NOT like the idea of setting the default encoding for Ubuntu AND windows back to UTF8.

Thanks for fixing this.

I have compiled the current "default" branch of Octave

• under (Ubuntu) linux, and
• with mxe-octave, and then using the resulting binaries under Windows 7.

These are the resulting encodings I see, when I save an m file from the gui editor (not changing the default settings):

• Ubuntu + Octave 4.0.1+ -> UTF-8
• Win 7 + Octave 4.0.1+ -> ISO8859-1

Under Windows 7 I have also loaded an m file with several umlauts (generated under Win7 + Octave 3.8.2 GUI) into the Octave 4.0.1+ editor, and the result look alright. And I've teste it vice versa, saving the umlauts in Octave 4.0.1+ and afterwards opening the file in Octave 3.8.2. Both looked alright.

So I would consider this issue to be resolved with this patch :)

On a Windows system the GUI Editor of Octave changed the code pages for the text encoding from version 3.8.2 to 4.0.0.

Here is how to reproduce this:

• use Windows (I had Win7)
• Create a new and empty document in the GUI edidtor of Octave.
• enter some special characters, for example umlauts like ÖÄÜ.
• save the small document as m file
• Have a look at the encoding of the generated m file

I observed the follwing encodings:

• Win7 + Octave 3.8.2-3 -> ANSI
• Win7 + Octave 4.0.0   -> UTF8
• Win7 + Matlab (2013b) -> ANSI
• Ubuntu + Octave 3.8.2 -> UTF8
• Ubuntu + Octave 4.0.0 -> UTF8

The resulting problem is: If you open an m file that was created by Matlab (or Octave 3.8.2) on a Windows OS with the current Octave 4.0.0, then all umlauts are displayed unreadable :(

This was working etter with (unofficial MXE-3) Windows release of Octave 3.8.2

Could be change the text encoding back to "ANSI" for the Windows release of the next Octave version?