Re: Coding systems and file completion for shell on w32

[Jason Rumney]

Lennart Borgman (gmail) wrote:
> We have previously on the developers list discussed coding systems for 
> *Shell* buffers on w32. I have sent some suggestions, but I do not 
> think  any changes have been done yet. Could we please try to fix this?

There are two parts to this. One part is to set the coding-system for 
input in default-coding-system based on the user's language environment, 
as is done on other platforms. The second is to consider whether we 
should make shell on Windows a special case and also set it's output 
coding-system. I disagree with the second part of this, as we do not 
know what commands the user will run, and what coding-system the output 
will be in. If the user is running commands like grep, diff, cat etc, 
then the coding-system will match the contents of the files they are 
processing. If they are mostly using dir, then the DOS codepage will 
certainly be better, but I do not think that is the most common use case.

What is needed to fix this is to have shell-mode set the coding system 
independantly for each command that is run, but I cannot think of a good 
way to detect reliably what coding system the output should be read as, 
unless we make a special exception for dir and any other commands we can 
think of that will cause output in the DOS codepage. Such a change would 
be for after the release.

> We have also discussed the tab file name completion in a cmdproxy 
> *Shell* buffer. This is currently broken. I sent a suggestion to fix 
> it, but I have got no feedback whatsoever on this.

Your suggestion must have been lost in the noise, that was a very long 
thread that deteriorated into off topic arguments so many people may 
have given up following it. Please send your suggestion again.