Package Details: shiboken2 5.15.16+3.1-2

Git Clone URL: https://aur.archlinux.org/pyside2.git (read-only, click to copy)
Package Base: pyside2
Description: Generates bindings for C++ libraries using CPython source code
Upstream URL: https://wiki.qt.io/Qt_for_Python
Licenses: LGPL-3.0-or-later
Submitter: arojas
Maintainer: envolution
Last Packager: envolution
Votes: 13
Popularity: 0.36
First Submitted: 2024-04-16 07:05 (UTC)
Last Updated: 2025-04-20 18:24 (UTC)

Latest Comments

« First ‹ Previous 1 2 3 4 5 6 7 Next › Last »

gyscos commented on 2025-06-30 18:10 (UTC) (edited on 2025-06-30 18:11 (UTC) by gyscos)

@gyscos Thanks - couple things - the only requirement is for make dependencies to be managed in the root package for all split packages. I guess runtime dependencies would be moved up to base if they were required for configuration/linking/inclusion purposes - but I can't think of a time I've had to do that. Make dependencies being pushed to 'depends' would only cause problems for users and won't solve anything in this situation.

At build time, all make + runtime dependencies must be installed, not just makedepends. It's pretty clear in the guidelines (https://man.archlinux.org/man/PKGBUILD.5#PACKAGE_SPLITTING) that the global depends + make depends need to include anything needed to build all packages. Essentially, if every split package overrides the global depends array, then that global depends array behaves exactly like the makedepends. It will not be present in the dependencies of any build package (since they all override that array), but is only required at build time.

I can see that python-numpy should be added to makedepends

qt5-base is also a dependency of one of the split packages (so it probably should be present at build time) but is not included in either make or runtime global depend.

envolution commented on 2025-06-30 18:05 (UTC)

@gyscos Thanks - couple things - the only requirement is for make dependencies to be managed in the root package for all split packages. I guess runtime dependencies would be moved up to base if they were required for configuration/linking/inclusion purposes - but I can't think of a time I've had to do that. Make dependencies being pushed to 'depends' would only cause problems for users and won't solve anything in this situation.

I can see that python-numpy should be added to makedepends - you can try install qt5-datavis3d, python-numpy, and makepkg -si from a clone of https://aur.archlinux.org/pyside2.git ... when reporting issues please use makepkg and not helpers.

I'll rebuild it in a short while to confirm things are working correctly with that change and make changes elsewhere if not

gyscos commented on 2025-06-30 17:44 (UTC) (edited on 2025-06-30 17:58 (UTC) by gyscos)

That message was initially copied from the output of makepkg -s, which cannot work anyway because qt5-datavis3d is in AUR.

The output from paru -B --noconfirm ., which is conceptually similar to makepkg -s, has a similar error message:

(... installs a bunch of dependencies ...)
fetching devel info...
==> Making package: pyside2 5.15.16+3.1-2 (Mon 30 Jun 2025 05:42:58 PM UTC)
==> Retrieving sources...
  -> Downloading pyside-setup-opensource-src-5.15.16.tar.xz...
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   315  100   315    0     0    662      0 --:--:-- --:--:-- --:--:--   663
100 3502k  100 3502k    0     0  5343k      0 --:--:-- --:--:-- --:--:-- 32.7M
  -> Downloading pyside2_5.15.16-3.1.debian.tar.xz...
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 28976  100 28976    0     0   716k      0 --:--:-- --:--:-- --:--:--  725k
  -> Found Modify-sendCommand-signatures.patch
  -> Found cmake-findpython.patch
==> Validating source files with sha256sums...
    pyside-setup-opensource-src-5.15.16.tar.xz ... Passed
    pyside2_5.15.16-3.1.debian.tar.xz ... Passed
    Modify-sendCommand-signatures.patch ... Passed
    cmake-findpython.patch ... Passed
==> Making package: pyside2 5.15.16+3.1-2 (Mon 30 Jun 2025 05:43:02 PM UTC)
==> Checking runtime dependencies...
==> Missing dependencies:
  -> qt5-multimedia
  -> qt5-tools
  -> qt5-sensors
  -> qt5-charts
  -> qt5-webengine
  -> qt5-datavis3d
  -> qt5-websockets
  -> qt5-speech
  -> qt5-3d
  -> qt5-svg
  -> qt5-script
  -> qt5-scxml
  -> qt5-x11extras
  -> qt5-quickcontrols2
  -> qt5-serialport
  -> qt5-remoteobjects
==> Checking buildtime dependencies...
==> ERROR: Could not resolve all dependencies.
error: failed to build 'pyside2-5.15.16+3.1-2 (shiboken2 pyside2 python-shiboken2 pyside2-tools)':
error: packages failed to build: pyside2-5.15.16+3.1-2 (shiboken2 pyside2 python-shiboken2 pyside2-tools)

But which package depends on qt5-datavis3d? I see pyside2 has it as opt-depend, but is it really required for building the package if no package requires it, and it's not a make-depend?

As mentioned in my previous message, if the global depends array is updated to properly reflect the state of each package depends array (or vice versa) as per the linked PKGBUILD guideline, then things build without issue.

envolution commented on 2025-06-30 17:39 (UTC)

@gyscos: could you share the few lines preceeding ==> ERROR: 'pacman' failed to install missing dependencies. ?

gyscos commented on 2025-06-30 15:05 (UTC) (edited on 2025-06-30 15:40 (UTC) by gyscos)

Hi, and thanks for maintaining this package!

It does seem that this PKGBUILD confuses makepkg. In a clean chroot, makepkg -s is not able to install the dependencies. And similarly, paru -B --noconfirm . is not able to install the dependencies by itself. In both cases the only way to build the package seems t be to manually install these qt5 dependencies, which indicates some PKGBUILD issue (the package should build without intervention in a clean chroot).

==> ERROR: 'pacman' failed to install missing dependencies.
==> Missing dependencies:
  -> qt5-multimedia
  -> qt5-tools
  -> qt5-sensors
  -> qt5-charts
  -> qt5-webengine
  -> qt5-datavis3d
  -> qt5-websockets
  -> qt5-speech
  -> qt5-3d
  -> qt5-svg
  -> qt5-script
  -> qt5-scxml
  -> qt5-x11extras
  -> qt5-quickcontrols2
  -> qt5-serialport
  -> qt5-remoteobjects
==> Checking buildtime dependencies...
==> ERROR: Could not resolve all dependencies.

For example, which package needs qt5-multimedia? It's not a make-depend, yet none of the resulting package depend on it.

Conversely, python-numpy is listed as a dependency of shiboken2 but is not included in the top-level depend or makedepend.

The PKGBUILD guideline states:

Note that makepkg does not consider split package depends when checking if dependencies are installed before package building and with --syncdeps. All packages required to make the package are required to be specified in the global depends and makedepends arrays.

Therefore, I think the top-level depend should be the union of the depends of all package (minus the internal packages this is building). If updated this way, this package would build properly with both makepkg and paru in a clean chroot:

diff --git a/PKGBUILD b/PKGBUILD
index 6f1c5e6..af68dff 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -16,9 +16,7 @@ pkgdesc="CPython bindings generator for C++ libraries - Debian rel ${pkgver#*+}"
 arch=(x86_64)
 url="https://wiki.qt.io/Qt_for_Python"
 license=('LGPL-3.0-or-later')
-depends=(qt5-multimedia qt5-tools qt5-sensors qt5-charts qt5-webengine qt5-datavis3d
-  qt5-websockets qt5-speech qt5-3d qt5-svg qt5-script qt5-scxml qt5-x11extras
-  qt5-quickcontrols2 qt5-serialport qt5-remoteobjects qt5-xmlpatterns)
+depends=(clang llvm libxslt qt5-xmlpatterns python python-numpy qt5-declarative qt5-base gcc-libs glibc)
 makedepends=(cmake python-setuptools python-wheel llvm clang)

 _debver="${pkgver/+/-}"

Alternatively, if these dependencies are actually required, they should be added to the respective package depends .

envolution commented on 2025-04-05 23:59 (UTC)

@echizenryoma try without that flag if you like, let me know if you find it to be faster or slower

echizenryoma commented on 2025-04-05 23:45 (UTC) 

cmake --build build --parallel 4

Why is parallel limited to 4 during build?

envolution commented on 2025-03-30 21:54 (UTC)

@19topgun93 Thanks for the report. cmake 4 update removed some methods used in the included cmake files - I've updated them with the new methods and included the patch in rel 2

19topgun93 commented on 2025-03-30 12:41 (UTC)

Hi I have problems with CMake.

=> Erstelle Paket: pyside2 5.15.16+3.1-1 (So 30 Mär 2025 14:32:20 CEST) ==> Prüfe Laufzeit-Abhängigkeiten... ==> Prüfe Buildtime-Abhängigkeiten... ==> WARNUNG: Verwende bestehenden $srcdir/ Baum ==> Entferne existierendes $pkgdir/ Verzeichnis... ==> Beginne build()... CMake Error at CMakeLists.txt:1 (cmake_minimum_required): Compatibility with CMake < 3.5 has been removed from CMake.

Update the VERSION argument <min> value. Or, use the <min>...<max> syntax to tell CMake that the project requires at least <min> but has been updated to work with policies introduced by <max> or earlier.

Or, add -DCMAKE_POLICY_VERSION_MINIMUM=3.5 to try configuring anyway.

-- Configuring incomplete, errors occurred! ==> FEHLER: Ein Fehler geschah in build(). Breche ab... -> Die Installationsschicht ist fehlgeschlagen, es wird zur nächsten Schicht übergegangen.error:Fehler beim Erstellen: pyside2 - exit status 4

GAthan commented on 2025-03-20 06:52 (UTC)

@envolution - thank you the pacman -S qt5-quickcontrols2 python-numpy worked a treat.

« First ‹ Previous 1 2 3 4 5 6 7 Next › Last »