Lennart Poettering

Well, doing "live" imports like that with journald isn't really the right tool for the job. journald is written with the job of collecting local logs in mind, it will...

classic fscrypt requires the decryption key to be available in the keychain of the process accessing the file system. It's a poor design decision if you ask me, because this...

So it appears this is fscrypt v2 policies that work without the keyring and they have been merged into the kernel. Hence, converting this into an RFE issue for homed.

It's just a matter of sitting down and doing the work. I think what to do is pretty straightforward. Patches would be very welcome.

I really think we should generate this server side, and then just display client side

So, in the long run we should add Varlink interfaces to relevant programs. journald for example has one now. or the userdb thing. It makes passing around JSON very natural,...

Hmm, for new Varlink interfaces we'd like to require they come with introspection data right-away, so that they can be introspected via the new "varlinkctl introspect" tool. Could you please...

i don't understand what this bug report is precisely about? That we don't send much details along with the polkit auth request for `org.freedesktop.systemd1.manage-units`? That's true, noone who looked into...

or hmm, varlink idl so far preferred prefix notation for arrays and such, i.e. it's `[]int`, and hence maybe we should also do the "+" as prefix? ``` method Foobar(x:...

hmm, and whether an upgrade is possible should probably be shown in a similar way. maybe: ```c method Foobar(x: bool) -> &(y:int) ``` or is all that too cryptic? ideas?