felix-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christian Schneider <ch...@die-schneider.net>
Subject Re: [Discuss] How should felix framework + gogo shell behave when Ctrl-C is received
Date Thu, 23 Mar 2017 14:36:15 GMT
I think what karaf assumes is that it is something like a shell. So when 
you press Ctrl-C in karaf and there is a foreground karaf command like 
tail -f running then it is ended. Not sure if this assumption is correct 
but as a user it feels right.

Ctrl-D ends the felix framework in the same way as in karaf. Would you 
propose to use Ctrl-D differently?


On 23.03.2017 15:05, Raymond Auge wrote:
> I really hate the current behaviour and have been wanting to fix it for
> years.
> However, Ctrl-C is a specific POSIX terminal code SIGINT which should, by
> default, terminate _foreground_ applications (things you invoked in the
> shell session).
> Meanwhile, in order to be consistent with shells and terminal emulators
> across the *nix-verse the proper _exit_ code should be EOF (i.e. Ctrl-D).
> I think both of these codes are not properly used in gogo.
> - Ray
> On Thu, Mar 23, 2017 at 9:17 AM, Christian Schneider <
> chris@die-schneider.net> wrote:
>> While working on documenting the gogo shell better (
>> https://issues.apache.org/jira/browse/FELIX-5595) I used the current
>> felix framework distro and tested a lot of the commands.
>> When I tried the log:tail command using -f to follow changes I noticed a
>> problematic behaviour. The tail -f command relies on the user pressing
>> Ctrl-C to interrupt the tail and go back to the shell. In Apache Karaf this
>> works like expected.
>> In the felix framework distro pressing Ctrl-C always exists the framework.
>> So it is impossible to stop a tail -f once you started it.
>> I opened an issue for this (https://issues.apache.org/jir
>> a/browse/FELIX-5597) as I think it would be better to have Ctrl-C behave
>> like in karaf.
>> I discussed this with Guillaume Nodet on irc and he generally agrees but
>> he asked me to bring this up on the list as the current behaviour might be
>> desired.
>> So what do you think? Should we try to change the behaviour or is it good
>> and expected that Ctrl-C ends the framework? In the latter case we would
>> need to have a different solution for ending a tail -f.
>> Christian
>> --
>> Christian Schneider
>> http://www.liquid-reality.de
>> Open Source Architect
>> http://www.talend.com

Christian Schneider

Open Source Architect

View raw message