I am using Konsole for a terminal session. It has "Default character encoding" set to UTF-8. In the session, $LANG=en_US.UTF-8. Yet when I cat a file with UTF-8 characters I get the question-mark inside an inverse diamond. Why don't I see the UTF-8 character? I have the Konsole font set to DejaVu Sans Mono 13pt.

I use xdce4-terminal but I too set it to DejaVu Sans Mono. A quick check shows that it can cat a file with runes in it, even though that font doesn't include runes.

The problem could be either specific to Konsole or it could be a wider KDE problem — not seeking an alternative font when the default is unsuitable. I can't solve either problem, but I can suggest an alternative terminal: mlterm is specifically designed to cope with any language. Of course that may not help if it's a KDE problem.

DavidMcCann: Thanks. I just tried mlterm. As installed, it didn't work either to show UTF-8 characters with 'cat'. I didn't see any way to change settings in mlterm, such as BG/FG colors, font or charset. Is there a way to do that? Maybe I need to config a charset setting or something.

Ah ha! Problem solved. Being new to the bewildering world of UTF-8 notation I had the wrong Hex bytes. My file had 00 B0, which is the "Hex Code Point" which, I suppose is represented as U+00B0. But that's not the Hex byte values. Those are C2 B0. When I changed the file to have C2 B0, then Konsole displayed the UTF-8 character correctly (degree symbol). The browser, however, display the degree symbol correctly whether or not the Hex characters were 00 B0 or C2 B0.

I'm sure there are all kinds of good reason for representing the hex sequence C2 B0 as U+00B0, but it confused me for a while.

ref: https://www.cogsci.ed.ac.uk/~richard...=00B0&mode=hex

A feature of browsers is that they try to be helpful and correct mistakes, which can make them a poor tool for debugging.


It may have helped to use cat -A and/or od to see actual bytes involved.

Also, Wikipedia's UTF-8 article has an example of how multi-byte encoding works.