k0001

I'd like to add that fonts such as **DejaVu Sans Mono** or **FreeMono** already support a wide variety of such mathematical symbols, operators, and letters.

Greek support will be excellent. Certainly, this is hard work. In any case, maybe it can be helpful to start by prioritizing some of these symbols according to how often...

Yes. Thanks for asking that. It seems that the most common mathematical symbols are, unsurprisingly, in the **Mathematical Operators** block: **U+2200–U+22FF**. There are some letter-like symbols, however, that fall outside...

@ChShersh I don't know of projects using `di` yet other than my private own. To be fair, though, this makes sense to some extent, seeing how the most recent version...

Actually, I think this can be made somewhat more backwards compatible. My previous comment was wrong: > By using conditional compilation (CPP) we could decide where to map > Int...

In https://github.com/tomjaguarpaw/haskell-opaleye/commit/9223756b187cee0f8c8931267aed59155a2250ce I made the change I mentioned in my last comment.

I should note that we could have an even more backwards compatible solution, albeit a bit more sophisticated: ``` haskell class ToPGInt8 a where pgInt8 :: a -> PGInt8 instance...

@tomjaguarpaw well, the current situation is beyond just “strange”: it will lead to loss on information on 64-bit systems. I for one have already added a wrapper around all this...

`fail = error` is the current situation, see https://hackage.haskell.org/package/base-4.11.1.0/docs/src/GHC.Base.html#fail (modulo stack trace details). Not being able to pattern match when using do notation makes using QuickCheck less ergonomic. This feature...