Sergey,
I've enabled c11 for parsec plugin at first. Then it ran in a pile of
problems with threads.h availability an ended up in removing c11 support.
https://github.com/mariadb-corporation/mariadb-connector-c/commit/b827743c27063e9f614dba4c731719240060e4ad
Still, even with c99 we can have our
Hi Nikita,
Back, when I was writing the original e-mail I didn't realize that we're
still on C99.
Yes, we'll definitely have to have different interfaces. I wonder if we're
considering
moving towards C11. Then we could probably try disabling our
pthread_yield() and
see if thrd_yield() is available
One concern could be that dropping our sched_yield wrappers, or
worsening its maintenance would affect the C part of the server, but I just
checked that there is a C11 function as well:
https://en.cppreference.com/w/c/thread/thrd_yield
I remember, though, that some systems don't supply threads.h,
Just want to add, that I think it's fine to have two different interfaces
for communicating with OS: one for C and another one for C++, just as we
use new(mem_root) in C++ and we have alloc_root for C. Same for atomics.
And if we'll have to support more (older) platforms, the library stuff is
eas
Steve Shaw writes:
> pipeline as other testing is currently underway so will probably be early
> next week.
Ok, sounds great!
> To confirm we will test the mariadb-11.4.6 branch as a baseline and then
I would suggest to use the commit e176066a9e02cf2bdedf9e79fb150029dde430f6
(which is the comm
Steve Shaw writes:
> One follow-up question, with the new build if I set log_bin=1 (or
> skip-log-bin=0) I get the following:
>
> 2025-05-16 15:17:43 0 [ERROR] Binlog name can not be set with
> --log-bin when --binlog-storage-engine is used. Use --binlog-directory
> to specify a separate director
