| 8 Mar 2026 |
K900 | As pulled in by opencv | 15:35:39 |
K900 | OK that's upstream's fault cool cool | 15:42:22 |
Gaétan Lepage | Great.
And that's the latest version? | 17:40:56 |
K900 | No, but latest is also broken | 17:41:17 |
whispers [& it/fae] | dotlambda put up a revert to 1.x | 17:41:57 |
whispers [& it/fae] | https://github.com/NixOS/nixpkgs/pull/497944 | 17:42:20 |
K900 | It's not a revert, it's monorepo nonsense | 17:43:04 |
K900 | But I guess if relaxDeps works then whatever | 17:44:01 |
K900 | Going to test all of my stuff again with those two PRs applied once they make it through the queue | 17:45:06 |
K900 | BUT WAIT THERE'S MORE OPENBLAS NONSENSE | 17:51:37 |
K900 | https://github.com/OpenMathLib/OpenBLAS/commit/d2906e8787ccc50051505f97262027bae6b55258 | 17:53:02 |
K900 | This time it's aarch64 AND single thread specific openblas nonsense | 17:53:15 |
K900 | OK nope anything down the chain from there also needs relaxDeps I guess | 18:02:58 |
Gaétan Lepage | I'll open a PR | 18:03:41 |
K900 | Oh wait no | 18:04:24 |
K900 | You can't | 18:04:25 |
K900 | It explodes | 18:04:28 |
K900 | But maybe if I update proto-plus | 18:05:24 |
K900 | https://github.com/NixOS/nixpkgs/pull/497977 | 18:16:42 |
K900 | OK this at least gets me to Celery | 18:16:45 |
K900 | Also, unixobcd | 19:24:41 |
K900 | https://github.com/NixOS/nixpkgs/pull/497996 | 19:24:41 |
K900 | This is going to staging-not-next because that would be too many rebuilds for a typo | 19:25:00 |
hexa | apache-orc also broken on protobuf | 19:57:12 |
hexa | [ 43%] Building CXX object c++/src/CMakeFiles/orc.dir/Timezone.cc.o
[ 44%] Building CXX object c++/src/CMakeFiles/orc.dir/Writer.cc.o
/build/source/c++/src/ColumnWriter.cc: In member function 'virtual void orc::ColumnWriter::writeIndex(std::vector<orc::proto::Stream>&) const':
/build/source/c++/src/ColumnWriter.cc:215:40: error: ignoring return value of 'bool google::protobuf::MessageLite::SerializeToZeroCopyStream(google::protobuf::io::ZeroCopyOutputStream*) const', declared with attribute 'nodiscard' [-Wer>
215 | rowIndex->SerializeToZeroCopyStream(indexStream.get());
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~
In file included from /nix/store/m6q48f6svqnfbvyqkis60a6vw5ww0v57-protobuf-34.0/include/google/protobuf/generated_message_tctable_decl.h:24,
from /build/source/build/c++/src/orc_proto.pb.h:24,
from /build/source/c++/src/wrap/orc-proto-wrapper.hh:48,
from /build/source/c++/src/BloomFilter.hh:23,
from /build/source/c++/src/ColumnWriter.hh:24,
from /build/source/c++/src/ColumnWriter.cc:26:
/nix/store/m6q48f6svqnfbvyqkis60a6vw5ww0v57-protobuf-34.0/include/google/protobuf/message_lite.h:751:44: note: declared here
751 | PROTOBUF_FUTURE_ADD_EARLY_NODISCARD bool SerializeToZeroCopyStream(
| ^~~~~~~~~~~~~~~~~~~~~~~~~
At global scope:
cc1plus: note: unrecognized command-line option '-Wno-unknown-warning-option' may have been intended to silence earlier diagnostics
cc1plus: all warnings being treated as errors
make[2]: *** [c++/src/CMakeFiles/orc.dir/build.make:324: c++/src/CMakeFiles/orc.dir/ColumnWriter.cc.o] Error 1
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [CMakeFiles/Makefile2:328: c++/src/CMakeFiles/orc.dir/all] Error 2
make: *** [Makefile:166: all] Error 2
| 19:57:45 |
Gaétan Lepage | Does it have a newer version? | 21:57:47 |
hexa | yeah, but the update didn't work cleanly | 21:58:24 |
| SomeoneSerge (matrix works sometimes) changed their display name from SomeoneSerge (back on matrix) to SomeoneSerge (matrix works sometimes). | 23:36:15 |
dotlambda | https://github.com/NixOS/nixpkgs/pull/497965 | 23:48:52 |
| 9 Mar 2026 |
Gaétan Lepage | protobuf_25 fix on staging-next
https://github.com/NixOS/nixpkgs/pull/497545 | 00:30:45 |