tl;dr:
TYVM. This is all great.
I'll be uploading *with* your dh changes, fixed, shortly.
Niels Thykier writes ("Bug#1089303: chiark-utils: Supporting rootless builds by
default"):
> The final patch `dh-prototype.diff` is what I envisioned would be the
> final bit required to move to `dh`. Though for some reason it changes
> the generated `shlib:*` substvars for `chiark-utils-bin`. It is
> consistently introduced with that patch (and reverting the patch
> consistently restores the original behavior) but I see no logic in the
> `dh` patch that should change the generated substvar, so I am at a loss
> here.
Thanks for the patch, which is actually excellent, and also for the
thorough and cautious report. You give me great confidence in the
code, this way.
I investigated the shlibs issue. I think the cause is that with the
dh sequencer, dh_link happens before dh_shlibdeps. The result is that
the symlink /usr/bin/xacpi-simple is found by the loop in what is now
override_dh_shlibdeps. The link target, xbatmon-simple, has its
dependencies explicitly downgraded, but the link isn't liosted in the
case statement. Previously, dh_link ran later. The fix is to have
the loop call realpath and match on the resolved path.
> Note that `dh` patch also have delta caused by `dh_dwz` being run.
Temporarily overriding this allowed me to get a very minimal diff out
of diffoscope.
> I did not check it for other issues once I spotted the shlib one.
I have now done this.
Your patch, and my fixups (see above, plus a few .gitignore additions)
will be in sid shortly.
Regards,
Ian.
--
Ian Jackson <ijack...@chiark.greenend.org.uk> These opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
