On Sep 23,  4:44pm, chris@qwirx.com (Chris Wilson) wrote:
-- Subject: Re: Problems with libedit and international characters

| > Yes, it will have a problem with UTF-8 and multi-byte character sets. It 
| > is not too diffcutly to fix though. I have not bothered because there is 
| > not enough demand. If you want to do it though, you can look at the tcsh 
| > code (where the editline code was created from) and copy the changes.
| 
| Perhaps that code could be factored out of tcsh so that we don't have 
| outdated copies of it lying around? And tcsh could use libedit?

That was the plan that never materialized :-) Unfortunately never found
the time to do it (and tcsh has a lot of nasty hooks in the edit code).

| Anyway I'm less scared of the libedit source code than I was, so now I 
| won't be afraid to hack at it if/when this bug bites my backside :-) But 
| right now I'm not sure I know what to do, unless it's trivial to copy out 
| of tcsh. (I'd need a library or function for utf-8 reading/writing, I 
| guess, and I don't know what that would be).

man mbrtowc... and look how tcsh handles input... It is not that hard to
do.

| > You need to call setlocale(); this works:
| 
| You're absolutely right, now it works on glibc too, my bad!
| 
| If this is required to use editline, then perhaps it could be documented 
| somewhere and added to the fileman example?

I guess I can add it to the example but calling setlocale() is required by
all programs that want to use localization, not just ones that use editline.

christos