Matthieu Dorier

Can you point me to the specific test source? The margo test is pretty simple, so I don't know what changed in Mercury 2.2.0 that made it no longer pass.

I may not be looking at the right place, I looked here: https://github.com/mercury-hpc/mercury/blob/master/Testing/test_rpc.c but it seems that the only test that is expecting a HG_NOENTRY is one where the sender...

I'm confused, [here](https://github.com/mercury-hpc/mercury/blob/master/Testing/test_rpc.c#L804) shouldn't it be `hg_ret != HG_NOENTRY` rather than `hg_ret != HG_SUCCESS` ?

Ok there was a bug in my margo test, which is fixed now, and the test passes, so the problem is somewhere else. The margo test was meant to reproduce...

> > I'm confused, [here](https://github.com/mercury-hpc/mercury/blob/master/Testing/test_rpc.c#L804) shouldn't it be `hg_ret != HG_NOENTRY` rather than `hg_ret != HG_SUCCESS` ? > > ok yes we could do that, I think it would be...

I bisected commits from 2.1.0 to 2.2.0 to get to the one that makes the problem appear in my code, and it's this one: https://github.com/mercury-hpc/mercury/commit/6fed707d9f2584cb71bbbd8ac083eed9a4ff4408 I don't know exactly what...

Ah I figured it out! To reproduce the problem, I need to register an RPC with a NULL handler (as if I were a client), then forward the RPC to...

I did a reproducer using Margo [here](https://github.com/mochi-hpc/mochi-margo/blob/main/tests/unit-tests/margo-forward.c) and there is an issue in Margo to track the problem [here](https://github.com/mochi-hpc/mochi-margo/issues/202). To summarize: Sending an RPC to a destination that has associated...

> Wrt the RPC ID issue that you have, which ID do you receive on the server end when you send ID=43 ? still 43 or some other random value...

I don't know if Mercury is zeroing the buffer beyond the set offset before sending or when receiving, but here is an example of what happens: - client sets the...