On Thu, 6 Jun 1996, Ian Jackson wrote:
> Paul Schoenly writes ("gzip and dpkg problem"):
> ...
> > $ dpkg --install package_x.deb
> ...
> > gzip:stdout: Broken pipe
> > dpkg-deb: subprocess gzip -dc returned error exit status 1
> > dpkg: error processing package_x.deb (--install)
> > subprocess dpkg-deb --control returned error exit status 2
> > Errors were encountered... (and so on)
>
> This means that SIGPIPE was set to SIG_IGN when dpkg started. For
> reasons too complicated to explain here this means that dpkg can't do
> proper error trapping (it always gets an error indication, and can't
> tell whether it's really an error).
>
[ much stuff deleted ]
>
> Please do tell us what the problem is if you find out. Telling us
> which ways of logging in cause the problem and which don't will help
> us a lot :-). If you don't want to bother trying to install packages
> (and it sometimes works even if the bug is present), try:
> cat /dev/zero | true
> If the bug isn't present this produces just `Broken pipe' (assuming
> you're using bash as your shell). If the bug is present it will
> produce `cat: write error: Broken pipe'.
ok, I am pretty sure I can blame this on xsm. Whenever I try this
test in rxvt start by xsm or in rxvt started from that original rxvt,
it fails, if I start rxvt from a window manager (fvwm2) which is
itself started by xsm, the test does not fail. Strange. I guess fvwm resets
that handler correctly.
