Arch Linux User Repository

Version 2.8.0 is trying to install an unnecessary ~20GB of new packages (rocm) on my system with no AMD cards.

Maybe it is better to have separate packages for different GPU vendors?

@gugah cuda is in the makedepends so I don't think any build dependency is missing for 2.8.0. Maybe it is because that you need to logout and login to make cuda's changes to environment variables work. You can create this package in a clean chroot by running makechrootpkg or extra-x86_64-build, those commands are available in package devtools.

As for the #include <algorithm> problem, this package is built with gcc14 until 2.8.0 so 002-fix-include-algorithm.diff is introduced. As we using cuda, gcc14 is replaced by gcc13 so there is no need to use this patch until cuda is also using gcc14. But applying this patch and build with gcc13 seems fine, so this patch is not removed.

@ZhangHua, maybe a build dependency is missing for 2.8.0? I'm getting:

-- Looking for a CUDA compiler - NOTFOUND
CMake Warning at /home/gugah/.cache/paru/clone/gpt4all-chat/src/gpt4all-2.8.0/gpt4all-backend/CMakeLists.txt:71 (message):
  CUDA Toolkit not found.  To build without CUDA, use -DLLMODEL_CUDA=OFF.

Even though CUDA is installed. Btw, upstream is fixing the #include <algorithm> error and some other missing includes that fail only with gcc14.

edit: I needed to logout/login before building gpt4all-chat as mentioned by the maintainer.

For the record, after some recent update (somewhere in the last 7 days), I got a verbose error on gpt4all-chat-git while the compiler checked the environment and I was able to finally compile gpt4all-chat by editing the PKGBUILD and adding -DLLMODEL_CUDA=OFF to the compiler options (I have an AMD card).

@gugah I checked your patch and find it works! thank you so much for your help! I have created a new release so everyone using this package can benefit from this patch.

AUR does not support pull request, so if you have any improvement to the repository, please contact maintainers directly or just leave your patch in a comment.

@javalsai, I've also had the exact same error when building. I tried changing some configs in /etc/makepkg.conf without luck

edit: A quick search points to a missing #include <algorithm> in gpt4all-backend/llamamodel.cpp. I'll check if it compiles with a simple patch.

edit 2: I was able to patch the build with

diff --git a/PKGBUILD b/PKGBUILD
index 4aa976a..d7bfe9c 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -12,6 +12,7 @@ makedepends=("cmake" "shaderc" "vulkan-tools" "vulkan-headers")
 source=(
     "$pkgname-$pkgver.tar.gz::https://github.com/nomic-ai/gpt4all/archive/refs/tags/v$pkgver.tar.gz"
     "001-change-binary-name.diff"
+    "002-fix-include-algorithm.diff"
 )
 declare -rAg _modules_name_map=(
     [gpt4all-backend/llama.cpp-mainline]=https://github.com/nomic-ai/llama.cpp/archive/a3f03b7e793ee611c4918235d4532ee535a9530d.tar.gz
@@ -35,6 +36,7 @@ do
 done
 sha256sums=('6849bfa2956019a3f24e350984fe9114b0c6e71932665640f770549d20721243'
             'c9f1242ff0dfd7367387d5e7d228b808cdb7f6a0a368ba37e326afb21c603a44'
+            '33353c4d0d7a5da7862c4965cf4e69452dda68d2dca184c38208cd6d20746913'
             'b47b1d8154a99304a406d564dfaad6dc91332b8bccc4ef15f1b2d2cce332b84b'
             '2fef47fc74c8ccc32b33b8c83f9833b6a4c02e09da8d688abb6ee35167652ea9')

@@ -65,6 +67,7 @@ prepare() {
         fi
     done
     patch -Np1 -i ../001-change-binary-name.diff
+    patch -Np1 -i ../002-fix-include-algorithm.diff
 }
 build() {
     cmake -B build-chat -S "$srcdir/gpt4all-$pkgver/gpt4all-chat" \

002-fix-include-algorithm.diff is trivial:

diff --git a/gpt4all-backend/llamamodel.cpp b/gpt4all-backend/llamamodel.cpp
index e88ad9f..b35bbdf 100644
--- a/gpt4all-backend/llamamodel.cpp
+++ b/gpt4all-backend/llamamodel.cpp
@@ -1,6 +1,7 @@
 #define LLAMAMODEL_H_I_KNOW_WHAT_I_AM_DOING_WHEN_INCLUDING_THIS_FILE
 #include "llamamodel_impl.h"

+#include <algorithm>
 #include <cassert>
 #include <cmath>
 #include <cstdio>

I would PR this fix but I don't know where to @ZhangHua

@ZhangHua neither of those commands worked, both gave the same errors in the chroot environment, I think I read somewhere in the wiki that it reads config from /etc/makepkg.conf, so I'm starting to think it could be an issue with that, but afaik mine is pretty normal, nothing special to break on.

Anyways, I don't really need this package, so it's not a priority for me to get it working, just leaving it here as I didn't find any related issue, if nobody else reports this it might just be me...

@javalsai It is indeed an error but I completely have no idea about this, because I am not familiar with C++. Maybe you can clone this aur repo and use makechrootpkg or extra-x86_64-build to build this package in a clean chroot?

You can check https://wiki.archlinux.org/title/DeveloperWiki:Building_in_a_clean_chroot for more info about building packages in clean chroot.

I somehow manage to get a different compile error each time, yes, I have base-devel installed:

/home/javalsai/.cache/paru/clone/gpt4all-chat/src/gpt4all-2.7.5/gpt4all-backend/llamamodel.cpp:45:21: error: no matching function for call to ‘find(std::vector<const char*>::const_iterator, std::vector<const char*>::const_iterator, const std::string&)’
   45 |     return std::find(EMBEDDING_ARCHES.begin(), EMBEDDING_ARCHES.end(), arch) < EMBEDDING_ARCHES.end();

/home/javalsai/.cache/paru/clone/gpt4all-chat/src/gpt4all-2.7.5/gpt4all-backend/llamamodel.cpp:674:20: error: ‘find_if’ is not a member of ‘std’; did you mean ‘find’?
  674 |     auto it = std::find_if(specs, std::end(specs),

/home/javalsai/.cache/paru/clone/gpt4all-chat/src/gpt4all-2.7.5/gpt4all-backend/llamamodel.cpp:892:22: error: ‘transform’ is not a member of ‘std’
  892 |                 std::transform(embd, embd_end, embd, [mean](double f){ return f - mean; });

etc (seems to be same kind of errors tho)

@nicholasr-ITSulu Installing base-devel is required for using AUR. See https://wiki.archlinux.org/title/Arch_User_Repository#Getting_started for more info.