On Thu, 09 Jan 2025 at 16:13:33 +0100, Svante Signell wrote:
> dh_exec is already used in most of the *.install files: (a bug?)
>
> gvfs-backends.install:
> #!/usr/bin/dh-exec
> ...
> usr/share/dbus-1/services/org.gtk.vfs.GPhoto2VolumeMonitor.service [!hurd-any]
> usr/share/dbus-1/services/org.gtk.vfs.GoaVolumeMonitor.service [!hurd-any
> !i386]
> usr/share/dbus-1/services/org.gtk.vfs.MTPVolumeMonitor.service [!hurd-any]
> ...
>
> Still dh_missing errors out:
> dh_missing
> dh_missing: warning: usr/share/dbus-
> 1/services/org.gtk.vfs.GoaVolumeMonitor.service exists in debian/tmp but is
> not
> installed to anywhere
I think that was a copy/paste error in 93fc54a1 "Stop enabling GOA on i386",
and it should just be [!i386]. Probably the architecture restriction was
copy/pasted from the AFC volume monitor, which is (apparently) unavailable
on Hurd *and* disabled on i386.
Now fixed in the packaging git repo, although there will be no practical
effect until/unless other build issues (possibly further down the
dependency chain) can be solved.
In general, if you're enabling or disabling components per-architecture,
part of that change will be to make sure the architecture restrictions
in the .install files match up with the architectures where you (now)
expect a particular component to be built/installed.
smcv
