Herbert Valerio Riedel
Herbert Valerio Riedel
related: https://ghc.haskell.org/trac/ghc/ticket/10920#comment:3
> If this feature is a welcome addition to cabal-plan, I would be happy to provide a PR for it. Yes, please! :-)
That's because for pre-7.10 I avoid boilerplate via `Setup.hs`, which in turns depends on a specific Cabal API. With Cabal 1.24 we will finally be able to specify build-dependencies for...
@amigalemming btw, you could try to `ghc hide Cabal-1.22.4.0` (for pre GHC 7.10s) as a local workaround
@Blaisorblade this is a bit of a gray area. `custom-setup` was designed to be usable as a backward-compatible annotation that could be retrofitted (via `.cabal`-revisions) in `cabal-version: 1.10` cabal files...
@pmlodawski `UUID` is an abstract data-type, i.e. its constructor is hidden so that its internal implementation can change (e.g. it could e.g. be arch-dependent whether it uses Word32 or Word64)...
@pmlodawski ...are you aware `aeson` versions now have From/ToJSON instances for `UUID`? :-)
> Knowing nothing about what causes this issue, something to note is that `regex-tdfa` supplies `SPECIALIZE` pragmas for `String`, both strict and lazy `ByteString`, and `Seq Char`. Notably, it does...
@A1kmm this seem unnecessary...? What concrete problem does this fix? I haven't noticed any issues with `double-conversion` and GHC >= 8.0; and I don't see how this would be dependent...
@A1kmm I've tried to reproduce this on Debian 9 with https://downloads.haskell.org/~debian/ and everything works just fine. In fact, your "fix" would break `double-conversion` for everyone else (at the very least)...