- 23 Jun, 2022 5 commits
-
-
Caleb Maclennan authored
-
Caleb Maclennan authored
namcap-devel: update path following !7 See merge request !10
-
-
Caleb Maclennan authored
pydepends: ignore invalid .py files Closes #26 and #25 See merge request !9
-
-
- 17 May, 2022 4 commits
-
-
Caleb Maclennan authored
-
Caleb Maclennan authored
-
Caleb Maclennan authored
Avoid filename conflicts on case-insensitive filesystems See merge request !7
-
-
- 07 May, 2022 2 commits
-
-
Caleb Maclennan authored
pydepends: recognize extension modules with limited ABI (*.abi3.so) See merge request !6
-
-
- 01 Feb, 2022 1 commit
-
-
Caleb Maclennan authored
Ignore ELF/RELRO checks for .debug packages See merge request !3
-
- 31 Jan, 2022 1 commit
-
-
Leonidas Spyropoulos authored
-
- 09 Dec, 2021 1 commit
-
-
Filipe Laíns authored
MutableMapping moved to collections.abc module See merge request !2
-
- 03 Dec, 2021 1 commit
-
-
Leonidas Spyropoulos authored
https://docs.python.org/3.10/library/collections.abc.html Signed-off-by:
Leonidas Spyropoulos <artafinde@gmail.com>
-
- 03 Sep, 2021 2 commits
-
-
Konstantin Gizdov authored
Fix warning for OFL license See merge request pacman/namcap!1
-
-
- 30 Mar, 2020 1 commit
-
-
The 'bzr' package was replaced with 'breezy' in the repos. Update the VCS makedepends rule to report the correct package. Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 29 Mar, 2020 2 commits
-
-
Namcap errors when a PKGBUILD has no source array. Test if the PKGBUILD has a source array in the VCSMakedepends rule to avoid such errors. Fixes FS#65042 Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Include a rule to check for vulnerable RUNPATH ELF binary entries, which allow arbitrary code execution by loading shared libraries from an attacker controller path. Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 15 Jan, 2020 1 commit
-
-
Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 15 Dec, 2019 1 commit
-
-
Sven-Hendrik Haase authored
Signed-off-by:
Sven-Hendrik Haase <svenstaro@gmail.com>
-
- 20 Nov, 2019 2 commits
-
-
Kyle Keen authored
Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Adds support for the following compression methods introduced in pacman 5.2: - zstd - lz4 - lzip Signed-off-by: George Rawlinson <george at rawlinson.net.nz> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 23 Jun, 2019 5 commits
-
-
As far as I can tell, this was never used. In commit 9b56647a the parsepkgbuild script was switched from defining a PATH as a directory in /tmp, to actually passing in a PATH=/dummy to `env -i`... which makes sense as the defined PATH did not used to be used. However, even before that, it's not safe to use mkdir -p in order to define a restricted shell without access to the PATH, as it's not guaranteed to be owned by the creator. In any event, the time for mkdir -p is long past. Signed-off-by: Eli Schwartz <eschwartz at archlinux.org> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
It is now possible to use b2sums to verify file integrity. See https://git.archlinux.org/pacman.git/commit/?id=e03752e6adc86cbb4cb4f52a38f6e3e98cbe9dd5 Signed-off-by: Eli Schwartz <eschwartz at archlinux.org> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
When listing out the contents of a variable, use loops and references to the variables in order to reduce code and make it easier to extend. In the future, a new metadata type can be added simply by appending to meta_keys=() Signed-off-by: Eli Schwartz <eschwartz at archlinux.org> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Just as it is unnecessary to run certain commands in a post_upgrade script, if a hook already runs it, so too is it usually unnecessary to depend on the package that installs the hook. Sometimes these dependencies are left in even when the install script is removed, because people lack understanding in the true subtleties of hooks and the commands they run -- but the commands do not depend on the list of files which trigger them in order to function, and are fully operative when installed later. In fact, they will already be required by the software that uses the results. e.g. desktop-file-utils and shared-mime-info provide hooks to compile the mime and desktop file databases, but upon first installing the hook, a full database will already be compiled. The hook does not need to be triggered by the filepath, and thus does not need to be a dependency of every package that provides a desktop or mime file. Instead, desktop environments which need a mime/desktop file database will depend on the package and thereby ensure the database exists when needed. Signed-off-by: Eli Schwartz <eschwartz at archlinux.org> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Arch Linux now has pacman hooks that provide automatic handling for update-{desktop,mime}-database without each package requiring a hard dependency and a post_upgrade hook. This means it is no longer correct to depend on the packages or provide an install script. Instead, remove this entirely. Fixes FS#56898 Signed-off-by: Eli Schwartz <eschwartz at archlinux.org> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 03 Feb, 2019 1 commit
-
-
Architecture-specific variables should be useable as soon as a package is not built for 'any' architecture. With the deprecation of i686, this becomes relevant, as many packages are now only built for x86_64. However, the source (and checksums) may be _not_ architecture-agnostic, e.g. should _not_ be listed as source= but source_x86_64=. This patch silences the extra-var-begins-without-underscore message which is currently emitted in that case. Signed-off-by: Erich Eckner <arch at eckner.net> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 02 Feb, 2019 1 commit
-
-
If a VCS source is renamed using the "::" syntax the makedepends are not detected. If there are files starting with <name of VCS binary> in the source array false positives are produced. See the gitlab package for example. Make the matching more robust to avoid such issues. Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 30 Nov, 2018 7 commits
-
-
Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Add a rule that warns about make dependencies already listed as dependencies. Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
For split pkgbuilds the pkgbase variable is optional. If not specified, the first element in the pkgname array is used. Currently parsepkgbuild.sh fails to detect split pkgbuilds if pkgbase is not specified. Therefore the split pkgbuild rules are not applied to such pkgbuilds. Instead of testing if pkgbase is set, test if the pkgname array has more than one element. If so, set pkgbase accordingly. Signed-off-by: Michael Straube <michael.straube at posteo.de> Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Use a PkgbuildRule instead of a PkgInfoRule Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
- 23 Sep, 2018 2 commits
-
-
Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-
Looks like DF_BIND_NOW has the same function as DT_BIND_NOW. Signed-off-by:
Kyle Keen <keenerd@gmail.com>
-