Bug#31521: dpkg: dpkg dying in eterm

[Ian Jackson <ian@chiark.greenend.org.uk>]

Daniel Jacobowitz writes ("Re: Bug#31521: dpkg: dpkg dying in eterm"):
> On Wed, Jan 06, 1999 at 06:44:46PM +0000, Ian Jackson wrote:
> > rather than
> >   jura:~> dd if=/dev/zero | true
> >   dd: /dev/full: Broken pipe
>         ^^^^^^^^^
> /dev/full ?

Err, oops :-).  The `transcript' there was mocked-up, because I
couldn't be bothered to reproduce the real bug conditions.

Here is a genuine transcript:
  jura:~> dd if=/dev/zero | true
  dd: standard output: Broken pipe
  9+0 records in
  8+0 records out
  jura:~>

Joey Hess writes ("Re: Bug#31521: dpkg: dpkg dying in eterm"):
> Ian Jackson wrote:
> > In case dpkg is too difficult a program, try the rune:
> >   dd if=/dev/zero | true
> > which should say
> >   jura:~> dd if=/dev/zero | true
> >   1+0 records in
> >   0+0 records out
> >   Broken pipe
> >   jura:~>
> > rather than
> >   jura:~> dd if=/dev/zero | true
> >   dd: /dev/full: Broken pipe
> >   1+0 records in
> >   0+0 records out
> >   jura:~>
> 
> I can verify the latter output in an xterm.

Presumably actually with `standard output' in the error message.

This means that either 1. xterm is buggy or 2. (more likely) xterm
inherited the brokennes from its parent.

You need to find the process whose descendants have the problem but
whose ancestors do not.  That process is where the bug is.

Fredrik Juhlin writes ("Re: Bug#31521: dpkg: dpkg dying in eterm"):
...
> However, trying it on my computer at home, which still runs Hamm and that
> has showed the same symptoms, I get:
> leonov:~# dd if=/dev/zero | true
> 9+0 records in
> 8+0 records out
> Broken pipe
> 
> I'm guessing that the important difference with the listings below
> is the dd:/dev/full... line, which doesn't show on any of the systems.

The important point is that the `Broken pipe' message is produced by
dd, rather than the shell.  dd puts `dd:' at the start.

> Btw, I tried removing and reinstalling packages (that formerly wouldn't
> get installed properly in an xterm) on my recently upgraded computer and
> it now works just fine.

The bug is probably fixed on your system now then.

...
> If you think a strace-log of a failed package-installation would help,
> I've got one lying around here somewhere.

No, thanks.  I know what is causing dpkg to misbehave and it's this
SIGPIPE problem.  That's not dpkg's fault.

Ian.