-
Notifications
You must be signed in to change notification settings - Fork 561
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
[OSRM] New Binary #6106
[OSRM] New Binary #6106
Conversation
@mattwigway Hey! Once this PR gets to ✅ and OSRM tags a new release, this should be the (note, at the moment only packaging the executables, do you also need the lib's?) |
@jeremiahpslewis that's great! Yes, I need the libs as well. Let me know if there's anything I can do to help. I've been working on creating a JLL myself that's nearly done, perhaps we can combine efforts. |
Gladly, perhaps this PR is a good starting point for a collaboration? Do you have any idea how to get the osrm build to output dynamically linked libraries? |
@mattwigway Just gave you access to the branch / fork so you can make changes. |
@jeremiahpslewis thanks, I'll take a look. Allegedly |
I also found that using the toolchain file for CMake avoided me needing to specify all the library locations: https://github.com/JuliaPackaging/BinaryBuilder.jl/blob/master/docs/src/build_tips.md#cmake-builds |
Attempting a build with the shared libraries now. |
I just ran it locally with your -D command, seems to work, so I've added the library products as well. :) |
It's possible the error I was running into is apple M1 specific - will see momentarily |
I was thinking it might also make sense to add the profiles (car.lua and friends) as FileProducts since OSRM is not very useful without these. |
@mattwigway Apple M1 is brilliant for Julia but not so good for BinaryBuild-ing. Have you tried using Yggdrasil via GitHub codespaces? It's all wired up so relatively painless. :) |
Sure, fine by me. Want to add it to the PR yourself? |
Will do. |
I'm building on x86, but the aarch64 mac target is failing. I think this is an upstream issue though. I'm trying to build OSRM outside of BinaryBuilder on an actual M1 Mac, and if that fails too I'll file an issue with the OSRM project. I think if that's the case we should just remove aarch64-apple-darwin as a supported platform from the jll for now. |
Yep, same error when building natively on M1. |
@mattwigway ok, so we are ✅ for linux, but windows and Mac seem to be blocked (upstream?) Release plan:
|
Thanks! I'll look quickly at the Windows build situation as it would be
ideal if I had a Windows binary available (I know, I know, I hate Windows
too... but unfortunately for the project I'm working on my hands are tied).
…On Fri, Jan 20, 2023 at 7:30 AM Jeremiah ***@***.***> wrote:
@mattwigway <https://github.com/mattwigway> ok, so we are ✅ for linux,
but windows and Mac seem to be blocked (upstream?)
Release plan:
- Get a new OSRM release tagged Project-OSRM/osrm-backend#6516
<Project-OSRM/osrm-backend#6516>
- Merge this PR and ship binary for linux
- Once mac / windows builds are debugged, then add and ship them
—
Reply to this email directly, view it on GitHub
<#6106 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAEKNLR4RNMN7SE2XNXEI6TWTKAPNANCNFSM6AAAAAAT7UWXBM>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
No luck getting the Windows binaries built - endless errors with mingw redefining GCC internal functions, and TBB not finding things from |
@mattwigway the latest oneTBB is also a disaster for windows, so until someone upstream makes this happen, think we‘ll have to stay content with Linux (and Mac someday?) do you know anyone you could ping to get a new version tagged of osrm? |
|
Mac (aarch64 error); x86_64 apple has the same error when the SDK is upgraded to 10.15:
...
|
|
Thanks, will update! |
Mingw error (x86-64-w64-mingw32-cxx03), which also pops up on
|
* Upgrade enzyme to refs/tags/v0.0.120 * Update build_tarballs.jl --------- Co-authored-by: wsmoses <[email protected]> Co-authored-by: William Moses <[email protected]>
Co-authored-by: Mosè Giordano <[email protected]>
* [ReferenceBLAS] Fix the suffix of xerbla_array
* [LAPACK] Update symbols to add missing suffix _64 * Add missing symbols
* Upgrade enzyme to refs/tags/v0.0.121 * Update build_tarballs.jl --------- Co-authored-by: wsmoses <[email protected]> Co-authored-by: William Moses <[email protected]>
Co-authored-by: github-actions <[email protected]>
Adds unordered set types
* Upgrade to v3.6.1 to hopefully get shtns_set_batch * Found more recent v3.6.6 on the Gitlab of Grenoble University * Try with SHTns v3.6.5 * Fix path to build directory * Upgrade to v3.6.6 --------- Co-authored-by: Thomas Dubos <[email protected]>
* add build_script * add missing scripts * update * fix typo * Update A/ADOLC/build_tarballs.jl Co-authored-by: Max Horn <[email protected]> * Update A/ADOLC/build_tarballs.jl Co-authored-by: Max Horn <[email protected]> * Re-introduce libjulia platforms * change platforms * Updated platforms. * Exclude musl due to compile error in ADOLC * Exclude Julia 1.11 due to (probably) pending adaptation of CxxWrap.jl * Update for Julia 1.11. * Update A/ADOLC/build_tarballs.jl ADOLC: chdir to source dir in one step Co-authored-by: Mosè Giordano <[email protected]> * ADOLC: use {prefix} instead of {WORKSPACE}/destdir Co-authored-by: Mosè Giordano <[email protected]> * ADOLC: remove superfluous empty line Co-authored-by: Mosè Giordano <[email protected]> * ADOLC: further streamlining * Remove version restrictions on llvm, libjulia, libcxxwrap_julia * Make libjulia, libcxxwrap_julia BinaryDependencies * ADOLC: add `clang_use_lld` = false Co-authored-by: Mosè Giordano <[email protected]> * ADOLC: reinstate compat bound for libcxxjulia, make it Dependency again * Bump Julia compat to 1.9+ * ADOLC: add expand_cxxstring_abis Co-authored-by: Mosè Giordano <[email protected]> --------- Co-authored-by: Max Horn <[email protected]> Co-authored-by: Jürgen Fuhrmann <[email protected]> Co-authored-by: Mosè Giordano <[email protected]>
* build LightGBM v4.3.0 with OpenCL/CUDA support * fix: replace FindCuda with FindCUDAToolkit * 😶 oops * patch switch to findcudatoolkit * boost pin compat * Update build_tarballs.jl * fix: add libs to CMAKE_CUDA_FLAGS * Update L/LightGBM/build_tarballs.jl --------- Co-authored-by: Mosè Giordano <[email protected]>
* [libunwind] Add v1.8.0 * [libunwind 1.8.0] Amend patch * [libunwind 1.8.0] Set preferred GCC version to 6 * [libunwind 1.8.0] idk, how about gcc 12 * [libunwind] 1.8.0 -> 1.8.1 * [[email protected]] Try `-fomit-frame-pointer` on Linux AArch64 Seems people building TensorFlow on Linux AArch64 hit the same inline assembly issue we're seeing here and this (plus `-flax-vector-conversion`) seems to fix it for them. This flag seems sus for libunwind but worth a try. * Thank you, Mosè Co-authored-by: Mosè Giordano <[email protected]> * [[email protected]] rm `-fomit-frame-pointer`, prefer GCC 12 Also add line breaks for long line * [[email protected]] Include patch for PR 748 Should hopefully fix the inline assembly issue on AArch64? * [[email protected]] Try re-lowering the preferred GCC version 12 seems excessive; 1.7.2 uses 5, let's try that * Update L/LibUnwind/[email protected]/build_tarballs.jl * Revert "Update L/LibUnwind/[email protected]/build_tarballs.jl" This reverts commit c37f8f9. --------- Co-authored-by: Mosè Giordano <[email protected]> Co-authored-by: Mosè Giordano <[email protected]>
Also fix GAC_LDFLAGS on macOS, drop a redundant filter!
The previous release had a bug with the `LBT_USE_RTLD_DEEPBIND` environment variable
* Upgrade enzyme to refs/tags/v0.0.122 * Update build_tarballs.jl --------- Co-authored-by: wsmoses <[email protected]> Co-authored-by: William Moses <[email protected]>
Release plan: