Re: [Haskell-cafe] Cabal message problem.

Daniel Fischer writes:

Or: it breaks with a bug introduced in 3.4.6 which hasn't yet been fixed.

This is an important point, I think: API breakages are not always intentional. Except for base, I generally don't specify upper bounds (well, maybe this is laziness on my part as well), unless I know it will break. On the other hand, I run a script that at regular intervals pulls a set of packages off Hackage and builds them, reporting any errors. This way, I'll hopefully be among the first to know if somebody upgrades a library that breaks any of my programs. -k -- If I haven't seen further, it is by standing in the footprints of giants