-
Notifications
You must be signed in to change notification settings - Fork 86
This issue was moved to a discussion.
You can continue the conversation there. Go to discussion →
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
You need to improve your google-rating #1736
Comments
Hi, thanks for your interest. We try to increase our popularities mostly by posting on social medias, e.g. Twitter, hacker news, etc. But looks like attractions come at a relatively low speed. Would be happy to know if you have any suggestions. Also feel free to tell us if you have any feature requests. |
good strategy - maybe it takes some time
builds under Linux with some errors on Fedora38 (mostly latest devtools per default) clean Fedora38 VM with
after install with standard cmake version 3.26.4 for Windows-builds - its a little bit unclear how to build for windows, what compiler is supported: build-instructions for CMake/VStudio or MSYS2/mingw64 would be nice i tried using MSYS2/mingw64 but that gives me this error after invoking cmake and ninja
using cmake+VS2022 generator gives me many include error for apache arrow while building, the linux build seems to download the dependency automaticly |
Hi @LowLevelMahn. For Fedora 38, the error seems to come from Node.js API. If you would like to build Node.is API, you need to run |
Build instructions specific to windows do seem to be missing. So far we've only managed to get it building with MSVC:
I'll add this info to the build section of README. I think there were other errors that I'd run into when I tried with mingw, but that error I think just means you didn't build arrow first (see below). Feel free to try mingw; I hadn't spent too much time digging into the issues there.
Are you using the Makefile on Windows? The same workaround is being done for both platforms in there. Unfortunately a cmake-only build doesn't work very well at the moment. |
is there a reason for using CMake from Make - why not use CMake exclusively? |
that worked - also needed to now it builds under linux with gcc 13.x and clang 16.x :) |
You can use CMake exclusively, but then you have to manually do what the Makefile does. Providing a Makefile simplifies the build instructions. I think you could do the same thing with the following commands (the directories used below are the ones from the makefile, and note that unless you manually pass mkdir external/build
cd external/build
cmake -DCMAKE_BUILD_TYPE=Release -G Ninja ..
cmake --build . --config Release -- -j %NUM_THREADS%
cd ../..
mkdir build/release
cd build/release
cmake -DCMAKE_BUILD_TYPE=Release -G Ninja ../..
cmake --build . --config Release -- -j %NUM_THREADS% We could probably add this as windows bat script or something to make things easier, but it would be a lot more work to port and maintain all the commands from the makefile. |
CMake is the multi-platform replacement for Make(and Autoconf/configure...) and CMake is also able to download files, install stuff, run sub-cmakes, make processes etc. but in a "more" portable way a proper configure CMakeLists.txt can usually replace a pre-Make step completely - so that building with Linux, MacOS, MSVC or MSYS2/gcc could work better out of the box - without any dependency to Make before the CMake run another question is: does the kuzu buildsystem fully support out-of-source builds - so that i can have my build folder anywhere i want? btw: the Apache Arrow build page is veryg good: what packages are needed on different OSes, how to install, win/mac/linux/msys2 build instructions, etc. |
So basically just using And actually, trying it again, I think it should be possible to use FetchContent to build arrow. It's just FetchContent with FindPackage that doesn't work (apache/arrow#32234). On the other hand, there seems to be some generated headers, so while the build worked for me when testing (with the exception of a linking error on the python library that can probably be fixed), it may not always since you have to call target_include_directories on both the in-source and generated headers and the generated headers may not be built before the code that includes them (which would be essentially the same issue I mentioned before).
Yes, but again, it's a little complicated right now. I've created a PR which includes making the Rust build be fully out-of-source (#1741), and the process is more or less what I outlined in my previous comment, including setting |
should work
better a little complicated right now then not possible :) |
The discussion here is becoming inactive for a while. I'm turning this into discussion for now. |
This issue was moved to a discussion.
You can continue the conversation there. Go to discussion →
you're feature-set seems to fullfill all my dreams - but google buries you deep behind all these Neo4J,Memgraph,etc. links :(
The text was updated successfully, but these errors were encountered: