Re: [elinks-dev] Bug#337159: elinks: Does not recover from CTRL-Z
Resending with a X-PTS-Approved header. On 08/01/23 23:28 +0200, Kalle Olavi Niemitalo said ... > Y Giridhar Appaji Nag <[EMAIL PROTECTED]> writes: > > > See also: http://bugs.debian.org/331409#57 > > Thank you for the pointer. However, I think it would be wrong > to reopen bug 331409 for the Bash-ELinks interaction, because > it was originally about a busy loop in Bash and that's not what > happens in bug 337159. > > What is the procedure for asking the Bash maintainer whether I am Cc:ing the bash PTS for a comment. > he considers the loss of SIGCONT a bug in Bash or in ELinks > (and if the latter, how else should ELinks be implemented)? Hi Matthias, can you please comment on this? From http://bugs.debian.org/337159#30 : The reason is that "fg" in Bash nowadays does not send SIGCONT to the job if Bash has seen from waitpid() that the job is already running. ELinks expects to receive this SIGCONT, in order to know when it should try to take control of the terminal again. This worked correctly in Bash 2.05 but has been broken since 2.05a. > I suppose I shouldn't just reassign bug 337159 to bash. We can do that at a later point. Giridhar -- Y Giridhar Appaji Nag | http://www.appaji.net/ signature.asc Description: Digital signature ___ elinks-dev mailing list elinks-dev@linuxfromscratch.org http://linuxfromscratch.org/mailman/listinfo/elinks-dev
Re: [elinks-dev] Bug#337159: elinks: Does not recover from CTRL-Z
On 08/01/23 23:28 +0200, Kalle Olavi Niemitalo said ... > Y Giridhar Appaji Nag <[EMAIL PROTECTED]> writes: > > > See also: http://bugs.debian.org/331409#57 > > Thank you for the pointer. However, I think it would be wrong > to reopen bug 331409 for the Bash-ELinks interaction, because > it was originally about a busy loop in Bash and that's not what > happens in bug 337159. > > What is the procedure for asking the Bash maintainer whether I am Cc:ing the bash PTS for a comment. > he considers the loss of SIGCONT a bug in Bash or in ELinks > (and if the latter, how else should ELinks be implemented)? Hi Matthias, can you please comment on this? From http://bugs.debian.org/337159#30 : The reason is that "fg" in Bash nowadays does not send SIGCONT to the job if Bash has seen from waitpid() that the job is already running. ELinks expects to receive this SIGCONT, in order to know when it should try to take control of the terminal again. This worked correctly in Bash 2.05 but has been broken since 2.05a. > I suppose I shouldn't just reassign bug 337159 to bash. We can do that at a later point. Giridhar -- Y Giridhar Appaji Nag | http://www.appaji.net/ signature.asc Description: Digital signature ___ elinks-dev mailing list elinks-dev@linuxfromscratch.org http://linuxfromscratch.org/mailman/listinfo/elinks-dev
Re: [elinks-dev] Bug#337159: elinks: Does not recover from CTRL-Z
Y Giridhar Appaji Nag <[EMAIL PROTECTED]> writes: > See also: http://bugs.debian.org/331409#57 Thank you for the pointer. However, I think it would be wrong to reopen bug 331409 for the Bash-ELinks interaction, because it was originally about a busy loop in Bash and that's not what happens in bug 337159. What is the procedure for asking the Bash maintainer whether he considers the loss of SIGCONT a bug in Bash or in ELinks (and if the latter, how else should ELinks be implemented)? I suppose I shouldn't just reassign bug 337159 to bash. pgp5TaeuBGrlu.pgp Description: PGP signature ___ elinks-dev mailing list elinks-dev@linuxfromscratch.org http://linuxfromscratch.org/mailman/listinfo/elinks-dev
Re: [elinks-dev] Bug#337159: elinks: Does not recover from CTRL-Z
On 08/01/23 10:47 +0200, Kalle Olavi Niemitalo said ... > Kalle Olavi Niemitalo <[EMAIL PROTECTED]> writes: > > > This worked correctly in Bash 2.05 but has been broken since 2.05a. > > These shells are also OK: > > csh 20070713-1 > pdksh 5.2.14-21 > tcsh 6.14.00-7 > zsh 4.0.6-18 > zsh 4.3.4-dev-7-2 See also: http://bugs.debian.org/331409#57 Giridhar -- Y Giridhar Appaji Nag | http://www.appaji.net/ signature.asc Description: Digital signature ___ elinks-dev mailing list elinks-dev@linuxfromscratch.org http://linuxfromscratch.org/mailman/listinfo/elinks-dev
Re: [elinks-dev] Bug#337159: elinks: Does not recover from CTRL-Z
Kalle Olavi Niemitalo <[EMAIL PROTECTED]> writes: > This worked correctly in Bash 2.05 but has been broken since 2.05a. These shells are also OK: csh 20070713-1 pdksh 5.2.14-21 tcsh 6.14.00-7 zsh 4.0.6-18 zsh 4.3.4-dev-7-2 pgpB8J6iHxDrK.pgp Description: PGP signature ___ elinks-dev mailing list elinks-dev@linuxfromscratch.org http://linuxfromscratch.org/mailman/listinfo/elinks-dev
Re: [elinks-dev] Bug#337159: elinks: Does not recover from CTRL-Z
Qingning Huo <[EMAIL PROTECTED]> writes: > So what is the real reason of this bug, elinks > is somehow confused whether it is at foreground? The reason is that "fg" in Bash nowadays does not send SIGCONT to the job if Bash has seen from waitpid() that the job is already running. ELinks expects to receive this SIGCONT, in order to know when it should try to take control of the terminal again. This worked correctly in Bash 2.05 but has been broken since 2.05a. The following patch does not revert any change between 2.05 and 2.05a but fixes or works around the bug anyway. diff -up /var/tmp/Kalle/Debian/bash-3.1dfsg/bash/jobs.c.\~1\~ /var/tmp/Kalle/Debian/bash-3.1dfsg/bash/jobs.c --- /var/tmp/Kalle/Debian/bash-3.1dfsg/bash/jobs.c.~1~ 2008-01-23 00:16:33.0 +0200 +++ /var/tmp/Kalle/Debian/bash-3.1dfsg/bash/jobs.c 2008-01-23 00:17:55.0 +0200 @@ -2754,8 +2754,7 @@ start_job (job, foreground) else jobs[job]->flags &= ~J_FOREGROUND; - /* If the job is already running, then don't bother jump-starting it. */ - if (already_running == 0) + if (1) { jobs[job]->flags |= J_NOTIFIED; killpg (jobs[job]->pgrp, SIGCONT); pgpIG1btv7Zeo.pgp Description: PGP signature ___ elinks-dev mailing list elinks-dev@linuxfromscratch.org http://linuxfromscratch.org/mailman/listinfo/elinks-dev