Bug 126007

Summary: European chars interpreted as prefixes and emacs loops on console
Product: [Fedora] Fedora Reporter: Eddahbi Karim <thetemplar>
Component: emacsAssignee: Jens Petersen <petersen>
Severity: high Docs Contact:
Priority: medium    
Version: rawhideCC: pawsa, sundaram
Target Milestone: ---   
Target Release: ---   
Hardware: i386   
OS: Linux   
Fixed In Version: 21.3-24 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2005-09-05 00:25:43 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---
Description Flags
emacs-kbd-coding.patch none

Description Eddahbi Karim 2004-06-15 08:13:20 UTC
Description of problem:
The French character î makes emacs freeze when using it in a real
terminal (CTRL+ALT+F1).

Version-Release number of selected component (if applicable):
GNU Emacs 21.3.1

How reproducible:
Always for me

Steps to Reproduce:
1. Use the fonts "latarcyrheb-sun16" with your terminal ($ setfont
2. Launch emacs then create a temporary text file ($ emacs test)
3. Type ^ i (Because you want to type a î)
Actual results:
Emacs will freeze and use 96% of CPU

Expected results:
Prints a î

Additional info:
It doesn't occur with the fonts lat9w-16 but with these fonts, 'î' is
taken like a macro-key (similar to C-, M-)

Comment 1 Jens Petersen 2004-09-29 16:15:34 UTC
"^i" works outside emacs?

Comment 2 Eddahbi Karim 2004-09-29 21:16:03 UTC
Yeah, it works. It just displays nothing but it works...

Comment 3 Jens Petersen 2004-10-06 07:38:33 UTC
Do I need to use a French kbd layout to reproduce this?

Can you explain exactly how to enter "^ i" in Emacs?

Comment 4 Jens Petersen 2004-10-06 08:15:18 UTC
Nevermind, I managed to input ^i with a French kbd layout.

Yep, I see it is interpreted as a prefix key by Emacs
but it doesn't emacs to freeze for me or use all the cpu.
Nor do the other accented letters.  However I'm testing
with a FC devel (rawhide).  Any chance you could test with
that? :)

Comment 5 Eddahbi Karim 2004-10-07 07:42:23 UTC
Did you have the same keyboard layout, fonts (latarcyrheb-sun16) and
charset (fr_FR.UTF8) ?
Are you using a xterm or a real terminal ? It seems to occurs only in
real terminals.

The first time I got this problem, it was with a fresh installation of
Fedora Core 2, using a french setup and the problem still occured with
rawhide packets, last time I checked.

If I find a friend with a Fedora Core 2 installed, I'll ask him to try

I'll try with the latest Rawhide when I can.
Thanks for the help ;)

Comment 6 John Smith 2005-02-10 00:02:39 UTC
I have a problem which I think is related to this one, although
it has nothing to do with the linux console, the font or the

Launch emacs (under X), then type:
C-x 8 ' e f o o b a r

Well here it makes emacs freeze and eat 98% of cpu before you
can even finish to type foobar. (You can unfreeze it with C-g,

Comment 7 Pawel Salek 2005-02-22 10:39:18 UTC
I see similar problem when I ssh from OSX to FC3 and type one of Swedish
characters, eg "ä" (use dead keys to construct it). Ctrl-G will not stop the loop.

Comment 8 Jens Petersen 2005-02-24 10:17:58 UTC
Yes, I can reproduce î and "C-x 8 ' e" being interpreted as a modifiers.
And on the console causing looping, which C-g breaks out of it.

I can't reproduce with "emacs --no-site-file" though:
it seems it is caused by the (set-keyboard-coding-system 'utf-8)
setting in lang-coding-systems-init.el (see bug 106929) afaict.

Comment 9 Jens Petersen 2005-02-24 10:57:18 UTC
Worse, it seems (set-keyboard-coding-system 'utf-8)
is needed for Latin input in a terminal, but this
breaks input on the console.  So probably need to
condition this on the value of TERM I guess.

Comment 10 Jens Petersen 2005-02-24 13:22:43 UTC
Created attachment 111375 [details]

I believe this change basically fixes the problem.
Could you please test it and if it is good, I'll
update emacs with it.

Comment 11 Jens Petersen 2005-02-25 13:34:18 UTC
The above fix appears to 21.3-24.
Could you please confirm if this fixes the issue?

Comment 12 Pawel Salek 2005-02-25 19:50:43 UTC
I have tested the patch and it appears to fix the issue - thanks!

Comment 13 Rahul Sundaram 2005-09-05 00:25:43 UTC
Closing as per previous comment