Refactor error handling in mods graph and dataset #31
Conversation
The unused error variants `GraphError` and `GraphMutationError` of sophia::error::Error where removed. The variant `UnsupportedGraphName` is moved to the new error for `GraphAsDataset`. The new error provides the moved variant and wrapps underlying (mutation) errors.
Actually, it is used. GraphAsDataset does not wrap a G(raph) but a Borrow (which is the trait bound on H). So you can decide to wrap either an owned graph or a simple reference to a graph.
Not sure what makes you think that... I tried to keep rio-specific code in rio_common.rs and all the parsers that are now based on rio (nt, nq, turtle, trig, ntrig), but parser.rs and serializer.rs are meant to be agnostic. If they are not, this is a bug.
I won't be able to work on it in the next week or two, so you are free to go. That being said, may be it would make sense to wait until theses modules are refactored to expose proper traits (as you suggested in #25)... |
|
Apart from all that, I like what's in this PR. Do you want me to merge it now, or do you plan to include the symetrical work in |
|
Actually the work for |
|
Ok cool, merging at once. After all, I am working now on the parser module, so you might want to wait a little bit before touching it... |
Continuation of #8 .
Removes unnecessary error variants from
sophia::error::Errorand adds a new error for theGraphtoDatasetwrapper.Future work
At investigating
GraphAsDataset<G, H>I recognized that the second parameterHis used nowhere actually. I could remove this within this PR if you like or did I missed something?@pchampin It seems to me that the mods parser and serializer are currently adapted to the
riocrates. Is this work finished and I can continue refactoring the error handling or should I wait?