Closed bluca closed 2 years ago
However, how do we confer that information back to the caller? We have to use the pre-defined
org.freedesktop.DBus.Error.*
code, and we are only left with its own free-form field for additional information. Should be put our reverse-domain code there unmodified? This would allow callers to react to it. But it would misuse the field. Alternatively, we convert it into a free-form string as last step. Unsure what to do exactly. Comments?
Hmm, I think I would vote for the second option. Not as useful, but the alternative is quite hacky.
Ok, updated - I defined the errors as suggested, added enums and converters. PTAL!
@dvdhrm ping
another ping @teg @dvdhrm
I am on 8month parental leave until March. I will defer any non-bugfixes until then. If you consider this PR critical, let me know and I will try to find some time for it.
It's not super urgent, and if you are on parental leave you definitely have more important things going on :-)
I am on 8month parental leave until March. I will defer any non-bugfixes until then. If you consider this PR critical, let me know and I will try to find some time for it.
@dvdhrm ping, in case you are back to work
@dvdhrm one more monthly ping. Anything else I should change here? Thanks!
Rebased to fix merge conflict
I merged this manually. I did a follow-up commit which changes bus1_error_*
to controller_name_error_*
, to make sure we keep the namespacing. Additionally, I changed the signed type to unsigned
, making value 0
be invalid. This is mostly to stay in line with our coding-style.
Thanks a lot for the patches! Very much appreciated!
Thanks - FYI if you had left that as feedback, I'd have been happy to update the PR myself
Thanks - FYI if you had left that as feedback, I'd have been happy to update the PR myself
I know, I know. But you already patiently waited for so long, so I didn't want to nag too much on coding style.
Example:
1) ConditionPathExists=/tmp/marker fails (note: with change in systemd to return
skipped
) 2) AssertPathExists=/tmp/marker fails 3) service immediately exits with failure before taking ownership of busDoes this sound sensible?
Fixes https://github.com/bus1/dbus-broker/issues/269