Timo Heister

@RyanMoulday Can you please also rebase to the most current master? This should fix the nedelec tests. @bangerth There are some shortcomings in the current approach (no reuse of Copy...

Ryan, is this ready for a another round of reviews?

> I do think that it would be interesting to try out the TaskFlow pipeline implementation I started looking into the tf::Pipeline and I have to admit that it is...

> I think that the change here is still awkward You should tell the original author about the bad test design (printing from a worker task! edit: this was me,...

> es a workstream call so perhaps the issue is here somewhere. It does not fail on master, so I think this is related to your changes here. The test...

Agreed. When it passes on your branch it only takes a few seconds: https://ci.tjhei.info/job/dealii-serial/job/PR-17119/13/testReport/projectroot.tests.grid.intergrid_constraintsdebug/ Do we have a deadlock in this test? Does it pass every time you run it...

I can reproduce the hang. Here is the call stack: ``` (gdb) bt #0 syscall () at ../sysdeps/unix/sysv/linux/x86_64/syscall.S:38 #1 0x0000738a9fcd9671 in std::__atomic_futex_unsigned_base::_M_futex_wait_until(unsigned int*, unsigned int, bool, std::chrono::duration, std::chrono::duration) () from...

> Where you able to produce the hang just running the test on its own? I had to launch 10 or more at the same time to trigger the problem....

> I can resolve this hang by creating a new executor within each WorkStream call which runs the one taskflow and then is cleaned up. Interesting. This sounds more and...

When you say > more than one WorkStream is run at the same time You mean a single program that executes several Workstreams sequentially after each other (all called from...