Was any breakage caused by 9.14?

1

Did anyone’s build break when they updated to GHC 9.14? That is, was any non-trival (not just bound bumps) change required to upgrade Haskell code that worked with GHC 9.12 to GHC 9.14? I haven’t found any examples but it’s possible I haven’t been paying close enough attention.

For the last coupule of years I’ve been keeping a breakage inventory of such breakage. The amount of breakage has been declining over that period. Has it really hit zero?

If you know of some such breakage please let me know, here or by filing a new issue on the Tilapia repo.

8 Likes
2

My proarrow package doesn’t build at all in 9.14.1, caused by this issue: 9.14 regression with ($) (#26543) · Issues · Glasgow Haskell Compiler / GHC · GitLab

5 Likes
3

timezone-series doesn’t compile with the new time-1.15 since it uses some internal modules that were hidden in this new time version.

But I am not sure if time is included in tilapia.

1 Like
4

Thanks for the info about timezone-series. The GHC breakage inventory is only about things for which changes are required to work with a particular version of GHC.

Since time is downgradeable I don’t think this change counts for this purpose as a breaking change in GHC 9.14. For example, this works: cabal repl -w ghc-9.14 -b timezone-series.

It would be a breaking change if there was some other collection of constraints for a build (it would probably have to be a very complex collection) that force time to 1.15 and thereby break the dependency on timezone-series. So if we can find an example of something like that I can add it to the inventory.

5

Thanks for the info about proarrow. I have started the GHC 9.14 breakage inventory and added it:

2 Likes