Sven-Hendrik Haase pushed to branch main at Arch Linux / Packaging / Packages /
torchvision
Commits:
9676ca46 by Sven-Hendrik Haase at 2023-11-28T08:05:57+01:00
upgpkg: 0.16.1-2: Build against ffmpeg 6 (#1)
- - - - -
2 changed files:
- .SRCINFO
- PKGBUILD
Changes:
=====================================
.SRCINFO
=====================================
@@ -1,14 +1,14 @@
pkgbase = torchvision
pkgdesc = Datasets, transforms, and models specific to computer vision
pkgver = 0.16.1
- pkgrel = 1
+ pkgrel = 2
url = https://github.com/pytorch/vision
arch = x86_64
license = BSD
makedepends = cmake
makedepends = ninja
makedepends = cuda
- makedepends = ffmpeg4.4
+ makedepends = ffmpeg
makedepends = python-pytorch-opt-cuda
makedepends = python-setuptools
makedepends = qt5-base
@@ -19,12 +19,14 @@ pkgbase = torchvision
depends = python-requests
depends = python-scipy
depends = python-sympy
- optdepends = ffmpeg4.4: video reader backend (the recommended one with
better performance)
+ optdepends = ffmpeg: video reader backend (the recommended one with
better performance)
optdepends = python-pycocotools: support for MS-COCO dataset
source =
vision-0.16.1.tar.gz::https://github.com/pytorch/vision/archive/v0.16.1.tar.gz
+ source =
torchvision-ffmpeg6.patch::https://patch-diff.githubusercontent.com/raw/pytorch/vision/pull/8096.patch
source =
https://github.com/NVIDIA/DALI/raw/main/dali/operators/reader/loader/video/nvdecode/cuviddec.h
source =
https://github.com/NVIDIA/DALI/raw/main/dali/operators/reader/loader/video/nvdecode/nvcuvid.h
b2sums =
b32f3f2142af2645c514364e0e5632e05af7a3f28b05af8200b8fbdc4edd0ded8d8595f1bcae0e21cfa795e1d557ef41dd2f21a584d0f4fd67e0fb4661f4ee8a
+ b2sums =
d9320af6029932045b95043728853a80c99d27ff919dc43eb2cac185181cee8a5ccbb4657ec77d281ceed22f27b8cfed4e7a7d783eecd477569641fd75ce4e95
b2sums =
9ccff204a4e1e93340d8b12c2b1d17e01663c12957b4665c0043eccf76d507a7308745a5d9e4d89657840aaf8abf0aa8f51bd79d6e0d5dc57a376d54a754755a
b2sums =
7db5d621f3099bc5455f1faeb7f4c3575a9cf70153ba56a6efc6d67d0ef2ac5438f6e117e621c5ef35c239eb3bce3fe17ce160e6b7765e8203d67a7299085429
=====================================
PKGBUILD
=====================================
@@ -19,7 +19,7 @@ _pkgname=vision
pkgbase='torchvision'
pkgname=('torchvision' 'torchvision-cuda' 'python-torchvision'
'python-torchvision-cuda')
pkgver=0.16.1
-pkgrel=1
+pkgrel=2
pkgdesc='Datasets, transforms, and models specific to computer vision'
arch=('x86_64')
url='https://github.com/pytorch/vision'
@@ -33,39 +33,33 @@ depends=(
python-sympy
)
optdepends=(
- 'ffmpeg4.4: video reader backend (the recommended one with better
performance)'
+ 'ffmpeg: video reader backend (the recommended one with better performance)'
'python-pycocotools: support for MS-COCO dataset'
)
makedepends=(
cmake
ninja
cuda
- ffmpeg4.4
+ ffmpeg
python-pytorch-opt-cuda
python-setuptools
qt5-base
nvidia-utils
)
source=("${_pkgname}-${pkgver}.tar.gz::https://github.com/pytorch/vision/archive/v${pkgver}.tar.gz";
+
$pkgname-ffmpeg6.patch::https://patch-diff.githubusercontent.com/raw/pytorch/vision/pull/8096.patch
"https://github.com/NVIDIA/DALI/raw/main/dali/operators/reader/loader/video/nvdecode/cuviddec.h";
"https://github.com/NVIDIA/DALI/raw/main/dali/operators/reader/loader/video/nvdecode/nvcuvid.h";
)
b2sums=('b32f3f2142af2645c514364e0e5632e05af7a3f28b05af8200b8fbdc4edd0ded8d8595f1bcae0e21cfa795e1d557ef41dd2f21a584d0f4fd67e0fb4661f4ee8a'
+
'd9320af6029932045b95043728853a80c99d27ff919dc43eb2cac185181cee8a5ccbb4657ec77d281ceed22f27b8cfed4e7a7d783eecd477569641fd75ce4e95'
'9ccff204a4e1e93340d8b12c2b1d17e01663c12957b4665c0043eccf76d507a7308745a5d9e4d89657840aaf8abf0aa8f51bd79d6e0d5dc57a376d54a754755a'
'7db5d621f3099bc5455f1faeb7f4c3575a9cf70153ba56a6efc6d67d0ef2ac5438f6e117e621c5ef35c239eb3bce3fe17ce160e6b7765e8203d67a7299085429')
prepare() {
cd "${srcdir}/${_pkgname}-${pkgver}"
- # fix building with ffmpeg4.4 by manually setup include and lib dir
- # and remove other codes to find ffmpeg exe, as ffmpeg4.4 are only headers
and libs without ffmpeg cmd
- sed -e 's#ffmpeg_include_dir = os.path.join(ffmpeg_root,
"include")#ffmpeg_include_dir = "/usr/include/ffmpeg4.4"#' \
- -e 's#ffmpeg_library_dir = os.path.join(ffmpeg_root,
"lib")#ffmpeg_library_dir = "/usr/lib/ffmpeg4.4"#' \
- -e 's#has_ffmpeg = ffmpeg_exe is not None#has_ffmpeg = True#' \
- -e '/ffmpeg_exe/d' \
- -e '/ffmpeg_bin/d' \
- -e '/ffmpeg_root/d' \
- -i setup.py
+ patch -Np1 -i "${srcdir}"/$pkgname-ffmpeg6.patch
cp -a "${srcdir}/${_pkgname}-${pkgver}"
"${srcdir}/${_pkgname}-cuda-${pkgver}"
cp -a "${srcdir}/${_pkgname}-${pkgver}"
"${srcdir}/python-${_pkgname}-${pkgver}"
View it on GitLab:
https://gitlab.archlinux.org/archlinux/packaging/packages/torchvision/-/commit/9676ca465d3d9ae8e19fac04007833e73ef721a3
--
View it on GitLab:
https://gitlab.archlinux.org/archlinux/packaging/packages/torchvision/-/commit/9676ca465d3d9ae8e19fac04007833e73ef721a3
You're receiving this email because of your account on gitlab.archlinux.org.
