Re: Review of new/changed lintian tag descriptions
Sorry, no patch yet - I've found the git repository, but I ran out of
time before I could work out how to check things out of it.
Niels Thykier wrote:
> @@ -223,11 +223,15 @@ Tag: experimental-to-unstable-without-comment
[...]
> + "experimental", this version has a distribution of "unstable", and there's
> + apparently no comment about the change of distributions.
> + .
> + Lintian looks in this version's changelog entry for the phrase "to
> + unstable" or "to sid", with or without quotation marks around the
> + distribution name.
> + .
> + This may indicate a mistake in setting the distribution and accidentally
> + uploading to unstable a package intended for experimental.
Making a mistake in doing something accidentally sounds unlucky.
Maybe it would be better as
This may indicate a mistake in setting the distribution and an accidental
upload to unstable of a package intended for experimental.
> +Tag: xs-testsuite-header-in-debian-control
> +Severity: wishlist
> +Certainty: certain
> +Info: There is an XS-Testsuite field in the <tt>debian/control</tt> file. As
> + of dpkg 1.17.10, the XS- prefix is no longer necessary. dpkg now
> + recognizes this header and handles it correctly. As of dpkg 1.17.11 the
> + field is automatically added by dpkg-source with the "autopkgtest" value if
> + there is a non-empty <tt>debian/tests/control</tt> file present. Consider
> + either removing the XS- prefix for this field or removing the field
> + altogether if it contains just the "autopkgtest" value.
> +
s/the "autopkgtest" value/the value "autopkgtest"/, at least the first
time (as a general rule, "the string FOO" is introducing that name,
"the FOO string" is mentioning a string identified by its familiar
name).
> Tag: xc-package-type-in-debian-control
> Severity: pedantic
> Certainty: certain
> @@ -233,3 +244,11 @@ Certainty: possible
> Info: The restriction formula in Build-Profiles field includes an unknown build
> profile. The only allowed build profiles are "stage1", "stage2", "nocheck",
> "nodoc", "nobiarch" and "cross".
> +
> +Tag: multiline-architecture-field
> +Severity: important
> +Certainty: certain
> +Ref: policy 5.6.8
> +Info: The values of the Architecture field in debian/control must not
> + be separated by anything else than spaces, i.e. must be single line
> + and is not allowed to span multiple lines.
(I wouldn't use "else" there myself, but plenty of other people seem
happy with it.)
You go from plural values to singular agreement on "is"; and "single
line" needs either an article (if it's nominal) or a hyphen (if it's a
compound modifier).
Info: The values of the Architecture field in debian/control must not
be separated by anything else than spaces; that is, they must form a
single line and are not allowed to span multiple lines.
(Also slightly repetitive, but at least it's clear.)
[...]
> @@ -52,8 +51,50 @@ Info: The package contains D-Bus policy configuration that uses
> .
> are redundant with the system bus' default-deny policy, and have
> unintended effects on other services.
Hmm, I'm not meant to be reviewing this, but the possessive of "system
bus" is "system bus's" (or "of the system bus").
> + .
> + This check ignores rules of the form
> + .
> + <policy user="root">
> + <allow ... />
> + </policy>
> + .
> + which are commonly used for the "agent" pattern seen in services like
> + BlueZ and NetworkManager: a root-privileged daemon calls out to
> + one or more per-user user interface agent processes with no specific
> + name, so <tt>send_destination</tt> is not easily applicable.
> + However, such rules should still be made as specific as possible to
> + avoid undesired side-effects.
> Ref: https://bugs.freedesktop.org/show_bug.cgi?id=18961,http://lists.freedesktop.org/archives/dbus/2008-February/009401.html
> -Experimental: yes
(Do they really have "no specific name", or is it just that they have
no name that can be predicted and applied in this sort of rule? Maybe
s/specific/specifiable/?)
[...]
> +Tag: uses-no-compression-for-control-tarball
> +Severity: important
> +Certainty: certain
> +Ref: deb(5)
> +Info: The control portion of this binary package uses a non compressed
> + format. Although dpkg will support extracting such binary packages
> + since dpkg 1.10.24, creating them is not advised except in special
> + cases.
> + .
> + Except if data is non compressible use gzip for
> + maximum compatibility and speed, and xz for maximum compression ratio.
"Non" is a non-adjective. "Non-compress{ed,ible}", or maybe indeed
"uncompress{ed,ible}". And maybe a comma before "use gzip".
[...]
> +
> +Tag: uses-no-compression-for-data-tarball
Ditto
[...]
> @@ -658,6 +658,8 @@ Info: The indicated program run in a maintainer script has a prepended
> .
> If the path is required to test a program for existence, one of the suggested
> workarounds in the developer's reference can be used.
> + .
> + See particularly function pathfind() of devref.
> Ref: policy 6.1, devref 6.4
Make that
See particularly the function pathfind() in devref.
(The abbreviation "devref" is standard Debian jargon, but when it's
spelled it out in full I would prefer "the Developer Reference".)
[...]
> +Tag: dep5-file-paragraph-reference-header-paragraph
> +Severity: normal
> +Certainty: possible
> +Ref: https://www.debian.org/doc/packaging-manuals/copyright-format/1.0/
> +Info: The Files paragraph uses a reference to a license, which is
^
Adding a comma before an English defining relative clause turns it
into a merely descriptive relative clause.
> + only defined in the Header paragraph. The copyright specification
> + requires that the Files paragraph either contains the full license
> + itself or references a "stand-alone" License paragraph, and not the
> + Header paragraph
^
Missing final punctuation.
> +
> +Tag: dep5-copyright-license-name-not-unique
> +Severity: normal
> +Certainty: possible
> +Ref: https://www.debian.org/doc/packaging-manuals/copyright-format/1.0/
> +Info: This paragraph define an already defined license.
^
It "defines" it.
> + .
> + According to specification the short license name are guaranteed
> + to be unique within a single copyright file.
I suppose "unique" might be misunderstood to mean that there can only
be one short license name per file, rather than that there can be any
number but they all have to be different. That probably doesn't
matter; a more plausible misunderstanding is that saying that they are
"guaranteed" to be unique makes it sound as if it's saying I don't
need to take special steps to make them unique, rather than that they
*must* be unique for the file to be compliant with the spec. And then
there's also some more number confusion: either "the name is unique"
or "(the) nameS ARE unique". So:
According to specification the short license names are required
to be unique within a single copyright file.
[...]
> @@ -19,7 +19,7 @@ Certainty: possible
> Info: The package contains a symlink with a target that
> appears to be a "failed" wildcard expansion. Furthermore
> the target does not exists in the package or any of its
^
It does not "exist".
> - direct dependencies (built from the same source).
> + direct dependencies (built from the same source package).
>
[...]
> +# Imported from lintian4python (python/watch)
> +Tag: debian-watch-file-accesses-pypi-over-http
> +Severity: wishlist
> +Certainty: certain
> +Info: The watch file is pointing at PyPI, using a http URL, not https.
> + PyPI now has https enabled, you should be able to simply switch to
> + https.
The dominant pronunciation is "aitch-tee-tee-pee", so "an http URL".
I suppose pedantically it's "the http scheme", but never mind.
The second line has a run-on sentence - either promote the comma to a
semicolon or add a "so".
> +
> +Tag: debian-watch-file-unsupported-pypi-url
> +Severity: important
> +Certainty: certain
> +Ref: #776267, https://wiki.python.org/moin/PyPISimple
> +Info: The watch file specifies a PyPI URL which is not a supported API.
^,
Here it *should* be a descriptive rather than defining relative clause
(it's saying "and that's not supported" rather than "a URL which is
not an API").
[...]
--
JBR with qualifications in linguistics, experience as a Debian
sysadmin, and probably no clue about this particular package
Reply to: