Hartmut Kaiser

> Hi @hkaiser. I was looking through this issue and I also couldn't reproduce it. > > I'm not sure if I'm trying to reproduce it on the right place....

> Hello all! This might be a novice question, but is it at all possible that the initial warning is not being reported because the function defined in the file...

> > Please test against the master branch as we will not patch up the old V1.8.1 release. > > I think maybe this issue is already solved. I tested...

@isidorostsa just in case you were wondering... the Windows tests all fail with a segmentation fault (error code `0xc0000005`)

@Pansysk75 it looks like that the new scheduler doesn't properly work yet (as you suggested). the clang-12 and gcc-10 errors are most likely related.

Excellent detective work @Pansysk75! Is that true in case if we used the consumer/producer token API the queue provides as well?

> > Excellent detective work @Pansysk75! Is that true in case if we used the consumer/producer token API the queue provides as well? > > Thanks! Here's what the implementer...

> Also, I think this fix is needed (or sth equivalent): > > ```diff > task_data thrds(d.num_thread_); > - thrds.tasks_.reserve(max_num_to_steal); > > > #ifdef HPX_HAVE_THREAD_STEALING_COUNTS > + thrds.tasks_.reserve(max_num_to_steal); > thread_id_ref_type...

> > Also, I think this fix is needed (or sth equivalent): > > ```diff > > task_data thrds(d.num_thread_); > > - thrds.tasks_.reserve(max_num_to_steal); > > > > > > #ifdef...

I'm not sure if anybody I know uses clangd for their work, so I wouldn't know whom to ask. This is probably caused by HPX being configured by one compiler...