Hi!
Hi all,
On Thu, Aug 08, 2019 at 02:10:39PM -0300, Lisandro Damián Nicanor Pérez Meyer wrote:
> > I had to use dpkg --force-overwrite to work around it, which showed the
> > additional messages:
> >
> > Preparing to unpack .../qtbase5-dev_5.11.3+dfsg1-2+b1_i386.deb ...
> > Unpacking qtbase5-dev:i386 (5.11.3+dfsg1-2+b1) over (5.11.3+dfsg1-2) ...
> > dpkg: warning: overriding problem because --force enabled:
> > dpkg: warning: trying to overwrite shared '/usr/share/qt5/doc/global/template/images/Qt-logo.png', which is different from other instances of package qtbase5-dev:i386
> > dpkg: warning: overriding problem because --force enabled:
> > dpkg: warning: trying to overwrite shared '/usr/share/qt5/doc/global/template/images/ico_out.png', which is different from other instances of package qtbase5-dev:i386
> > dpkg: warning: overriding problem because --force enabled:
> > dpkg: warning: trying to overwrite shared '/usr/share/qt5/doc/global/template/style/list_arrow.png', which is different from other instances of package qtbase5-dev:i386
> > dpkg: warning: overriding problem because --force enabled:
> > dpkg: warning: trying to overwrite shared '/usr/share/qt5/doc/global/template/style/list_expand.png', which is different from other instances of package qtbase5-dev:i386
> >
> > Usually, for a multiarch issue, the +b1 in the version number would make me suspicious, but maybe not here, the image seems to contain a creation time.
>
> So we have two issues here:
>
> - On one side those images should not have a creation date due to
> reproducibility. Actually this has never happened before and buildds might
> have built everything in deifferent dates.
> - On the other I wonder why I left those files there. This comes from 2015,
> so I'll have to dig into it further.
The issue is caused by dh_strip_nondeterminism changing PNG metadata
timestamps to dates from binNMUs, which are different on amd64 and i386.
I have just filed a bug against debhelper and described this issue in
more detail, see https://bugs.debian.org/934405.
As a workaround, we should tell dh_strip_nondeterminism to ignore these
files using its -X option. I will be able to do that in a couple of days.
That's a part of it. I sincerely don't remember why I kept those files there instead than -doc.
