Klemens Böswirth

I can't find the issue now, but IIRC the decision was to make key names read-only based on a second reference counter that only counts references from `KeySet`s.

Yes, of course the simplest solution is to not lock/unlock. But then it is kinda hard to keep track whether a `Key` is contained in a `KeySet`, i.e. whether it...

It should be fixed in #3935. I'll try to finish the PR this week. The code is already there, if you wanna test it. Documentation and tests are still missing,...

I'd have to check again for the full details, but: - I know `toml` definitely doesn't (fully) support metadata. - https://github.com/ElektraInitiative/libelektra/blob/master/src/plugins/toml/toml/checks_toml_1_0_0/roundtrip.toml still contains some commented out test-cases with `TODOs` -...

I think if someone has prior experience with flex/bison parsing (not unreasonable, considering there are other courses for that), then working on this should be possible. The code is pretty...

#2330 is closed, did you mean #3910 or maybe #4466?

> We do not need to make such an attempt: We can internally do a cast. That's exactly what the standard is talking about. The object (`KeySet`) was defined as...

> I don't see how a cast does a modification of the object The cast doesn't the cast is what creates the "lvalue with non-const-qualified type". But adding the hashmap...

> Yes, but that would be in an different statement/expression and your quote would not apply anymore. I'm not sure what you want to say here. The way I read...

> Is this also the case if the object initially wasn't const? (As is always the case in our situation.) I'm not sure. "an object defined with a const-qualified type"...