Andrey Mokhov

@jmtd What if the user has already provided escaped strings, e.g. we have `\"x\"` in the input? Do we want to mess this up by changing it to `\\"x\\"`? Or...

@jmtd Thanks! I have a suspicion that this might be too simple and fail to handle this case correctly: ```haskell -- A string containing a single character " test1 ::...

My 'test framework' is not something I'm proud of and could definitely be improved (#13). However, for this PR it should be fairly straightforward to add tests somewhere around here:...

By the way, it may be useful to expose the escaping functionality for two reasons: * It would be much easier to document and test the escaping functionality in isolation....

@zyklotomic Hi there, thanks for taking this over! My preference would be to finish the `NonEmpty` PR first, and then implement SCC as a separate PR. It would just make...

I think it's fine to keep working on `NonEmpty` in this PR. If you could just undo the SCC-related changes for now that would be great. Of course, please take...

> Given a graph with different node types (say "a", "b", "c", etc.) and different edge types ("r", "s", "t"), is it acceptable to modify connect such that the edge...

> > Is there anything in alga that would help building such graphs? > > Not at the moment. Apart from the simplest case with two vertex types, of course....

@boggle Supporting what you describe in full generality is going to be hard, most likely impractical in a language like Haskell, because you'll have to encode these edge-vertex compatibility predicates...

> limit the set of valid expressions of the algebra instead of modifying the algebra itself I don't think I suggested that. By limiting the set of valid expressions we...