Re: Controling requirements when packaging haskell modules
On Thu, 17 May 2018 16:18:49 +0200 l...@gnu.org (Ludovic Courtès) wrote: ... > Ton, could you update your patch series at: > > https://bugs.gnu.org/31422 > https://bugs.gnu.org/31423 > > ? > > TIA! > > Ludo’. Will do, trying to finish and send them now. On a related note: I started packaging these to get haskells sdl2 bindings. But I can't figure out how to get through the problem of requirements, building ghc-linear currently gives me: Configuring linear-1.20.7... Warning: This package indirectly depends on multiple versions of the same package. This is highly likely to cause a compile failure. package lens-4.15.4 requires semigroupoids-5.1 package kan-extensions-5.0.1 requires semigroupoids-5.1 package free-4.12.4 requires semigroupoids-5.1 package adjunctions-4.3 requires semigroupoids-5.1 package linear-1.20.7 requires semigroupoids-5.2.2 phase `configure' succeeded after 1.6 seconds And then the build fails a little later because of what looks like a missing feature of semigroupoids-5.1... My code for both ghc-linear and ghc-sdl2 is available[1], but can't build because of this. Not sure I'll work more on solving it either, as I'm not going to use them myself. [1]:https://notabug.org/thomassgn/guixsd-configuration/src/master/modules/ton-haskell.scm#L59 pgp1YyTYaEdFv.pgp Description: OpenPGP digital signature
Re: Controling requirements when packaging haskell modules
Hello, Ricardo Wurmus skribis: > Ludovic Courtès writes: > >>> You should not package ghc-array, ghc-containers, or ghc-binary — these >>> are all provided by GHC itself. It is possible that there are some more >>> packages for which this applies (possibly including semigroupoids). >> >> Oops, I pushed ghc-array and ghc-containers just yesterday. Should I >> revert? > > Yes, these packages better should not exist. You can revert the commits > adding them and modify the package definitions of whatever new packages > refer to them. > > We removed them some time ago when we cleaned up and updated most of the > Haskell packages. Right. I’ve removed them again now, sorry for the mess. Ton, could you update your patch series at: https://bugs.gnu.org/31422 https://bugs.gnu.org/31423 ? TIA! Ludo’.
Re: Controling requirements when packaging haskell modules
Ludovic Courtès writes: >> You should not package ghc-array, ghc-containers, or ghc-binary — these >> are all provided by GHC itself. It is possible that there are some more >> packages for which this applies (possibly including semigroupoids). > > Oops, I pushed ghc-array and ghc-containers just yesterday. Should I > revert? Yes, these packages better should not exist. You can revert the commits adding them and modify the package definitions of whatever new packages refer to them. We removed them some time ago when we cleaned up and updated most of the Haskell packages. -- Ricardo
Re: Controling requirements when packaging haskell modules
On Mon, 14 May 2018 13:20:48 +0200 Ricardo Wurmus wrote: > Ludovic Courtès writes: > > >> You should not package ghc-array, ghc-containers, or ghc-binary — these > >> are all provided by GHC itself. It is possible that there are some more > >> packages for which this applies (possibly including semigroupoids). > > > > Oops, I pushed ghc-array and ghc-containers just yesterday. Should I > > revert? > > Yes, these packages better should not exist. You can revert the commits > adding them and modify the package definitions of whatever new packages > refer to them. > > We removed them some time ago when we cleaned up and updated most of the > Haskell packages. > > -- > Ricardo > > Hi, I saw this message yesterday and thought I'd change and fix my patches, but had other things at that time. I should have sent a short message to the bug containing the patches. I'm going to clean up and send new patches without the ghc-array and so on. (I tested several of the requirements/dependencies and I think I isolated what needs to be declared). pgpgjkBM_ZsLY.pgp Description: OpenPGP digital signature
Re: Controling requirements when packaging haskell modules
Hello, Ricardo Wurmus skribis: > Ton writes: > >> I'm trying to package ghc-sdl2, but have hit a problem with requirements I >> don't understand how to get past. >> >> The work so far can be seen and downloaded from >> https://notabug.org/thomassgn/guixsd-configuration/src/master/modules/ton-haskell.scm > > You should not package ghc-array, ghc-containers, or ghc-binary — these > are all provided by GHC itself. It is possible that there are some more > packages for which this applies (possibly including semigroupoids). Oops, I pushed ghc-array and ghc-containers just yesterday. Should I revert? Thanks, Ludo’.
Re: Controling requirements when packaging haskell modules
Ton writes: > I'm trying to package ghc-sdl2, but have hit a problem with requirements I > don't understand how to get past. > > The work so far can be seen and downloaded from > https://notabug.org/thomassgn/guixsd-configuration/src/master/modules/ton-haskell.scm You should not package ghc-array, ghc-containers, or ghc-binary — these are all provided by GHC itself. It is possible that there are some more packages for which this applies (possibly including semigroupoids). The next thing you should check is that you’re packaging only those versions of Haskell packages that are part of LTSHaskell or Stackage. Do not package the latest versions. -- Ricardo