This kind of file includes translated entries, which causes spelling
check false positives, and makes the author tests and CI systems
checking those fail.

Some of these are valid, but we'll fix them in the soon to open 1.22.x
series. This should make the CI systems based on Debian unstable
checking author mode, get back to normal.

Add the requested module for this Debian derivative. While ideally this
would all be parametrized in fields on the origin files, this is
currently not specified, so until that happens, it's convenient to ship
this in here.

Requested-by: Ivan J. <parazyd@dyne.org>

This should give cppcheck a better understanding of what the code
semantics are.

Bind it to the actual files currently triggering this report.

Keeping these line-based suppressions in sync with the code is
cumbersome and error-prone. Move them as inline suppressions in
comments, so that they can easily follow the line number when the
code in the file moves around.

Use the long version to make the option obvious, as there is no need for
succinctness in scripts, and no portability concerns.

Use the long version to make the option obvious, as there is no need for
succinctness in scripts, and no portability concerns.

Fixes: commit 6cad120a
Changelog: silent

Warned-by: codespell

This new hook will return a regex matching the version part that denotes
that this is a backport, with a capture group for the backport delimiter
string.

The multicd name does not properly describe what this method is supposed
to be able to handle, as we now have many other media types where
distributions can ship their packages and updates on.

Remove all current explicit signal handling code for window resizes
and instead rely on ncurses doing this for us, which is going to be
async-signal-safe, instead of the current code which was not.

Changelog: internal

These are all program that are part of the core dpkg instead of the
build tools. Move them with the rest of the run-time programs to apply
the same constraints and infrastructure.

- Move dpkg-deb and dpkg-split under src/ as these are part of the dpkg
  toolset.
- Move common source files under src/common/.
- Move program specific source files under subdirectories in src/.
- Rename the various *cmd.c into <tool>/main.c.
- Set automake option subdir-objects.

This unclutters the source tree, and moves the files where they belong
in normal autotools build systems.

This can be useful for other distributions and non-derivative systems
that package dpkg. It will also make it possible to add native systemd
timers and use a single shared implementation.

This new check has started triggering, which we were previously ignoring
with a different tag.

Warned-by: shellcheck

New shellcheck versions emit an error for sourced files that cannot be
found. Enable their parsing, and refer the sourced points to the current
files in the source tree.

Warned-by: shellcheck

Warned-by: cppcheck

This is triggering on many parts that might need to be handled later on,
but right now produce noise.

Warned-by: cppcheck

Changelog: internal
Warned-by: lgtm

The Perl version in Debian buster is 5.28.1, which is the release that
will be oldstable once 1.21.x gets uploaded to Debian unstable.

Fixes: commit bc140c0e
Changelog: silent

This program will be used by dpkg-maintscript-helper and any maintainter
script that needs a realpath that can handle pathnames relative to the
dpkg root directory.

We need to disable some more false positives related to cppcheck not
understanding non-returning functions.

Files that have been autogenerated are of no interest because in many
cases we have little direct control over them.

We need to disable some more false positives. And while we should be
configuring some -I and -D, we cannot, as cppcheck lack of understanding
of non-returning functions makes the subsequent analysis useless.

The new python version makes codespell itself emit runtime warnings,
which confuses the test suite. Ignore thse for now.

Some got introduced in 1.89 and fixed in 1.90, others had been fixed
before but never noticed, remove them now.

Update the comments to match the newly grouped suppressions.

This is a cppcheck false-positive, but by rearranging the code we can
avoid it, and it makes it somewhat more clear anyway.

This was causing lots of fuzzying when the line numbers were getting
changed.

Warned-by: cppcheck

This was really a redundant condition, and independent of the variable
being an enum instead of a bool.

Warned-by: cppcheck