This is necessary to keep proper track of arbitrary, additional columns. We also have a dedicated function to concatenate janno files, which makes use of combineTwoJannos.
Is is easy to forget to use these functions, though, because Data.List.concate will of course also typecheck. I almost ran into this issue yesterday.
Can we prevent this by making [JannoRow] an instance of Monoid? We should probably give it a proper name in the process, maybe JannoFile.
I think we only have to implement <>/mappend for Semigroup (which is essentially combineTwoJannos) and mempty which should be []. We could also overwrite the default mconcat (foldr mappend mempty) with the potentially more efficient foldl' mappend mempty, which is what we currently have in concatJannos.
I feel like we already discussed sth. like this in the past, @stschiff. I hope I didn't forget a crucial drawback of this idea.
A janno file is currently defined as
[JannoRow]. To combine two janno files we have a dedicated functionThis is necessary to keep proper track of arbitrary, additional columns. We also have a dedicated function to concatenate janno files, which makes use of
combineTwoJannos.Is is easy to forget to use these functions, though, because
Data.List.concatewill of course also typecheck. I almost ran into this issue yesterday.Can we prevent this by making
[JannoRow]an instance ofMonoid? We should probably give it a proper name in the process, maybeJannoFile.I think we only have to implement
<>/mappendforSemigroup(which is essentiallycombineTwoJannos) andmemptywhich should be[]. We could also overwrite the defaultmconcat(foldr mappend mempty) with the potentially more efficientfoldl' mappend mempty, which is what we currently have inconcatJannos.I feel like we already discussed sth. like this in the past, @stschiff. I hope I didn't forget a crucial drawback of this idea.