Investigate what's up with building codegate with llama-cpp-python 0.3.6

[jhrozek]

I don't have more details, but llama-cpp-python 0.3.6 broke our image builds. We just reverted the dep bump without investigating more, but we should to allow us to be up-to-date on the package.

[lukehinds]

interesting , I would be it was some missing c-library or version mismatch

[aponcedeleonch]

More context:

Strangely, building the image locally failed using make image-build but not on CI with our action image-build. The culprit was llama-cpp-python (0.3.6). Logs on failure below.

Steps to Reproduce

1. Change the version of llama_cpp_python in pyproject.toml from

llama_cpp_python = "==0.3.5"

to

llama_cpp_python = "==0.3.6"

2. Run locally make image-build

#0 25.77   
#0 25.77   -- Configuring done (0.6s)
#0 25.77   -- Generating done (0.0s)
#0 25.77   -- Build files have been written to: /tmp/tmp0d6vwh24/build
#0 25.77   *** Building project with Ninja...
#0 25.77   Change Dir: '/tmp/tmp0d6vwh24/build'
#0 25.77   
#0 25.77   Run Build Command(s): /tmp/tmpy46tdodh/.venv/bin/ninja -v
#0 25.77   [1/60] /usr/bin/g++ -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-array-bounds -Wextra-semi -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-threading.cpp
#0 25.77   [2/60] /usr/bin/gcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wdouble-promotion -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-alloc.c
#0 25.77   [3/60] /usr/bin/gcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -DGGML_USE_OPENMP -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/.. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-cpu -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wdouble-promotion -mcpu=native+nodotprod+noi8mm+nosve -fopenmp -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-cpu/ggml-cpu.c
#0 25.77   FAILED: vendor/llama.cpp/ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o 
#0 25.77   /usr/bin/gcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -DGGML_USE_OPENMP -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/.. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-cpu -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wdouble-promotion -mcpu=native+nodotprod+noi8mm+nosve -fopenmp -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-cpu/ggml-cpu.c
#0 25.77   cc1: error: unknown value 'native+nodotprod+noi8mm+nosve' for '-mcpu'
#0 25.77   cc1: note: valid arguments are: cortex-a34 cortex-a35 cortex-a53 cortex-a57 cortex-a72 cortex-a73 thunderx thunderxt88p1 thunderxt88 octeontx octeontx81 octeontx83 thunderxt81 thunderxt83 ampere1 emag xgene1 falkor qdf24xx exynos-m1 phecda thunderx2t99p1 vulcan thunderx2t99 cortex-a55 cortex-a75 cortex-a76 cortex-a76ae cortex-a77 cortex-a78 cortex-a78ae cortex-a78c cortex-a65 cortex-a65ae cortex-x1 ares neoverse-n1 neoverse-e1 octeontx2 octeontx2t98 octeontx2t96 octeontx2t93 octeontx2f95 octeontx2f95n octeontx2f95mm a64fx tsv110 thunderx3t110 zeus neoverse-v1 neoverse-512tvb saphira cortex-a57.cortex-a53 cortex-a72.cortex-a53 cortex-a73.cortex-a35 cortex-a73.cortex-a53 cortex-a75.cortex-a55 cortex-a76.cortex-a55 cortex-r82 cortex-a510 cortex-a710 cortex-x2 neoverse-n2 demeter generic
#0 25.77   [4/60] /usr/bin/g++ -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-array-bounds -Wextra-semi -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-backend.cpp
#0 25.77   [5/60] /usr/bin/g++ -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-array-bounds -Wextra-semi -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-opt.cpp
#0 25.77   [6/60] /usr/bin/gcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wdouble-promotion -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml.c
#0 25.77   [7/60] /usr/bin/g++ -DGGML_BACKEND_SHARED -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-array-bounds -Wextra-semi -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-backend-reg.cpp
#0 25.77   [8/60] /usr/bin/g++ -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-array-bounds -Wextra-semi -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/gguf.cpp
#0 25.77   [9/60] /usr/bin/gcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/. -I/tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/../include -O3 -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wdouble-promotion -MD -MT vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o -MF vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o.d -o vendor/llama.cpp/ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o -c /tmp/tmpqqg7v28w/llama_cpp_python-0.3.6/vendor/llama.cpp/ggml/src/ggml-quants.c
#0 25.77   ninja: build stopped: subcommand failed.
#0 25.77   
#0 25.77   
#0 25.77   *** CMake build failed
#0 25.77   
#0 25.77 
#0 25.77   at /usr/local/lib/python3.12/site-packages/poetry/installation/chef.py:164 in _prepare
#0 25.77       160│ 
#0 25.77       161│                 error = ChefBuildError("\n\n".join(message_parts))
#0 25.77       162│ 
#0 25.77       163│             if error is not None:
#0 25.77     → 164│                 raise error from None
#0 25.77       165│ 
#0 25.77       166│             return path
#0 25.77       167│ 
#0 25.77       168│     def _prepare_sdist(self, archive: Path, destination: Path | None = None) -> Path:
#0 25.77 
#0 25.77 Note: This error originates from the build backend, and is likely not a problem with poetry but with llama-cpp-python (0.3.6) not supporting PEP 517 builds. You can verify this by running 'pip wheel --no-cache-dir --use-pep517 "llama-cpp-python (==0.3.6)"'.
#0 25.77 
------
Dockerfile:18
--------------------
  17 |     # Configure Poetry and install dependencies
  18 | >>> RUN poetry config virtualenvs.create false && \
  19 | >>>     poetry install --no-dev
  20 |     
--------------------
ERROR: failed to solve: process "/bin/sh -c poetry config virtualenvs.create false &&     poetry install --no-dev" did not complete successfully: exit code: 1
make: *** [image-build] Error 1