to avoid build error due to conflicting declaration:
flow/Profiler.actor.cpp: In function 'uint64_t gettid()':
flow/Profiler.actor.cpp:56:17: error: ambiguating new declaration of 'uint64_t gettid()'
FILE* f;
^
In file included from /nix/store/4wy9j24psf9ny4di3anjs7yk2fvfb0gq-glibc-2.31-dev/include/unistd.h:1170:0,
from ./flow/Platform.h:49,
from ./flow/flow.h:40,
from flow/Profiler.actor.cpp:39:
/nix/store/4wy9j24psf9ny4di3anjs7yk2fvfb0gq-glibc-2.31-dev/include/bits/unistd_ext.h:34:16: note: old declaration '__pid_t gettid()'
extern __pid_t gettid (void) __THROW;
^~~~~~
Since glibc 2.28, `<sys/types.h>` no longer includes
`<sys/sysmacros.h>`, which provides these macros,
so we have to explicitly import the latter, too, to fix the following
build problems:
flow/Platform.cpp: In function 'void getDiskStatistics(const string&, uint64_t&, uint64_t&, uint64_t&, uint64_t&, uint64_t&, uint64_t&)':
flow/Platform.cpp:626:56: error: 'gnu_dev_major' was not declared in this scope
if(majorId == (unsigned int) gnu_dev_major(buf.st_dev) && minorId == (unsigned int) gnu_dev_minor(buf.st_dev)) {
^
flow/Platform.cpp:626:111: error: 'gnu_dev_minor' was not declared in this scope
if(majorId == (unsigned int) gnu_dev_major(buf.st_dev) && minorId == (unsigned int) gnu_dev_minor(buf.st_dev)) {
^
Enables testing in parallel. This cuts down the test phase from over
five to under one minute on my 6C/12T processor.
Sadly while the upstream runs their tests using `-n auto` this isn't
working for us in all cases. I currently assume they use low concurrency in
their CI, which is not triggering these issues..
All packages were outdated.
Asterisk 15 is not supported anymore, but there is 17 now.
All versions bumped pjproject to 2.10 which requires overriding the
prefix.
Since Asterisk 17, `make install-headers` seems to be needed.
This reverts commit c45160366b
and hopefully fixes#99236: evaluation on Hydra.
For lack of a better method, I used the live jobset to bisect.
On the reverted commit it failed (tried several times because sometimes
evaluation fail transiently, possibly due to exhausting some resources)
and on the parent it succeeded: https://hydra.nixos.org/eval/1617490
Right now I don't have energy to find what exactly is wrong in the
commit, and it doesn't seem important in comparison to nixos-unstable
channel being stuck on a commit over one week old.