João Meira

Does #210 solve this issue? It does not allow plugins to store config values in `mptcpd.conf`, but it allows plugins to have their own configuration file.

> Several of the tests in the GitHub build are also failing with these changes in places (see the "Checks" tab). Yeah, I noticed. Since I didn't know if you...

Although it did not appear during the checks, `test-cxx-build` actually fails with `ell` >= 0.45. In `ell` version 0.45 `DEFINE_CLEANUP_FUNC` was added, and since `ell/settings.h` is included in the `mptcpd/types.h`...

That's very interesting. I use the same, gcc 11.2. I tested it on my Arch linux, in one system with Archlabs and also in a Fedora vm, all with gcc...

> Are you configuring mptcpd in a specific way or just running `configure` without any command line options or variable tweaks? Just `./configure`.

Would it be possible to bump the version of ell to one >= 0.45 of the automated builds to check if the error occurs?

I bumped the version of ell of the automated build in one of my forks and got the same error (check https://github.com/MPTCP-Lab/mptcpd/tree/plugins_confs_test2 which contains the patch without the workaround and...

The error only occurs with this patch, since it adds `#include ` to `mptcpd/types.h`. With the master version the error does not occur.

> Do and need to be included in ? Generally, we try to avoid directly or indirectly including ELL headers in the public headers. I have the habit of just...

Couldn't it also be used for defining the event receiving order for the functions that use `l_hashmap_foreach`, (_i.e._ `mptcpd_plugin_new_interface`, `mptcpd_plugin_update_interface`, `mptcpd_plugin_delete_interface`, `mptcpd_plugin_new_local_address` and `mptcpd_plugin_delete_local_address`)? At first this probably appears that...