Tags naming convention

[Guillem Jover <guillem@debian.org>]

Hi!

I've noticed the recent addition of tags with either very confusing
names like:

- debian-watch-may-check-gpg-signature

  As reported in #735040, where the suggested name seems way better:
  debian-watch-does-not-check-for-gpg-signature

- privacy-breach-may-use-debian-package

  Does the privacy breach use the package? A better name could perhaps be:
  privacy-breach-uses-embedded-file

Or names that instead of stating the detected fact, seem to dictate what
it would like to see.

- file-should-not-be-compressed

- debian-rules-should-not-automatically-update-control
- debian-rules-should-not-use-DEB_BUILD_OPTS
- debian-rules-should-not-use-or-modify-user-only-variable
- debian-rules-should-not-use-pwd
- debian-rules-should-not-use-underscore-variable

- maintainer-script-should-not-hide-init-failure
- maintainer-script-should-not-modify-ld-so-conf
- maintainer-script-should-not-modify-netbase-managed-file
- maintainer-script-should-not-use-adduser-system-without-home
- maintainer-script-should-not-use-ancient-dpkg-epoch-check
- maintainer-script-should-not-use-ancient-dpkg-multi-conrep-check
- maintainer-script-should-not-use-deprecated-chown-usage
- maintainer-script-should-not-use-dpkg-status-directly
- maintainer-script-should-not-use-fc-cache
- maintainer-script-should-not-use-gconftool
- maintainer-script-should-not-use-install-sgmlcatalog
- maintainer-script-should-not-use-service
- maintainer-script-should-not-use-start-stop-daemon
- maintainer-script-should-not-use-update-alternatives-remove
- maintainer-script-should-not-use-update-alternatives-set

I agree with Jakub Wilk's recent comments on the list that these are not
good names either. Lintian detects patterns, some might be problems that
must be fixed in all cases, others might be a matter of policy, others
might perhaps be issues sometimes, and that's why lintian allows
overridding/disabling them either per package or per profile. Encoding
either the severity/certainty or the possible solution in the tag name
duplicates the information contained elsewhere and makes them awkward
to change.

I'd request that no more such tag names be added, and ideally the
current ones be renamed, although the longer they stay the more
overrides they might accumulate. :/

I skimmed over other tag names and I've found also these patterns which
raise red flags for me (might have missed some), and do not conform
with the vast majority of other tags, or even related ones:

# -must-not-

- udeb-postinst-must-not-call-ldconfig

# -should-not-

- web-application-should-not-depend-unconditionally-on-apache2
- orphaned-package-should-not-have-uploaders
- changelog-should-not-mention-nmu
- debian-revision-should-not-be-zero
- library-in-debug-or-profile-should-not-be-stripped

# -might-not-

- description-synopsis-might-not-be-phrased-properly

# -should-

- changelog-should-mention-nmu
- changelog-should-mention-qa
- clean-should-be-satisfied-by-build-depends
- copyright-should-refer-to-common-license-file-for-apache-2
- copyright-should-refer-to-common-license-file-for-gfdl
- copyright-should-refer-to-common-license-file-for-gpl
- copyright-should-refer-to-common-license-file-for-lgpl
- debian-watch-file-should-dversionmangle-not-uversionmangle
- debian-watch-file-should-mangle-version
- debian-watch-file-should-use-sf-redirector
- debian-watch-file-should-uversionmangle-not-dversionmangle
- debug-file-should-use-detached-symbols
- debug-package-should-be-named-dbg
- debug-package-should-be-priority-extra
- games-package-should-be-section-games
- init.d-script-should-depend-on-virtual-facility
- menu-method-should-include-menu-h
- new-package-should-close-itp-bug
- postrm-should-call-ldconfig
- symlink-should-be-absolute
- symlink-should-be-relative
- transitional-package-should-be-oldlibs-extra

Thanks,
Guillem