Joseph Schuchart

I think @bosilca is right: I have two interfaces (one being the wifi and one being a docker device). If I a) run with `--mca btl_tcp_if_include 127.0.0.1/16` (notice the `/16`...

Here is my network config: ``` $ ifconfig -a docker0: flags=4099 mtu 1500 inet 172.17.0.1 netmask 255.255.0.0 broadcast 172.17.255.255 ether 02:42:34:cb:48:a7 txqueuelen 0 (Ethernet) RX packets 0 bytes 0 (0.0...

Removing the blocker label since https://github.com/open-mpi/ompi/pull/10349 was merged and the user-facing issue was fixed. The underlying discrepancy between info allocation and object semantics remains though and should be addressed.

We should leave it open for now. The fixes in https://github.com/open-mpi/ompi/pull/10420 and https://github.com/open-mpi/ompi/pull/10349 didn't fix the underlying issue of info objects not being pure ompi objects. We'll have to discuss...

I pushed a bunch of changes. The biggest change is that a continuation request is now derived from `ompi_request_t`, which shrunk `ompi_request_cont_data_t` considerably and cuts down on the pointer chasing....

Oh, and I will squash everything down to one commit once we're done...

@bosilca I pushed a few changes: 1) Now using the new `OMPI_MPIEXT_*_POST_CONFIG` hook. I hope this is how it is intended to be used. 2) I had to make sure...

@gkatev Sorry for the late reply, this slipped under my radar. As @bosilca mentioned, MPI (§7.4.4) requires us to drop any info keys that we don't know about (to provide...

I see, thanks for the clarification. `opal_infosubscribe_change_info` looks for subscribers for each key in the `info` object and invokes the subscriber callback. If there are no subscribers for a particular...

We don't have to remove (yet) unused keys in `MPI_Info_get`, just hide them. That way, if they are referenced later they will appear later. Still not great, as two calls...