#14558: Unable to parse integer-gmp's Cabal file -------------------------------------+------------------------------------- Reporter: taylorfausak | Owner: (none) Type: task | Status: new Priority: normal | Milestone: Component: Core Libraries | Version: 8.2.2 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Changes (by simonpj): * owner: hvr => (none) * status: closed => new * resolution: fixed => Comment: For me, an important principle is this: * '''GHC should strive to make life as easy as possible for downstream tools''' One way to make life easy is never to make any changes: eternal backward-compatibility. No one is asking for that. But once we leave that extreme there's a continuous spectrum of choices, and we have to strike a balance. For example, in the language design, and library changes, we go to much effort to avoid breaking downstream libraries and tools, except when there is no alternative -- and even then we try to offer plenty of notice and workarounds. I'm not familiar with all the technical issues on this thread, but I can see that there is at least a perception by some that we have caused unnecessary breakage --- breakage that I didn't even know GHC ''could'' cause. Far from making no concessions, I think the GHC developers should be very open to making concessions to better support GHC's user community. I'll have a chat with Ben, Simon, Herbert and others, to see how best to move forward. But meanwhile I wanted to state the principle explicitly. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14558#comment:48 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler