We picked up some extensions along the way in the dependency, so that went to None, then the things that depended on it devolved from SafeInferred to None as well.
-Edward On Mon, Oct 29, 2012 at 5:54 PM, Michael Sloan <mgsl...@gmail.com> wrote: > I guess that's a good way to safeguard against future accidental > toggleage. Still, it's puzzling that the status of "Safe-Infered" was > lost. > > On Mon, Oct 29, 2012 at 2:46 PM, Edward Kmett <ekm...@gmail.com> wrote: > > I fixed it. Version 3.0.6 was just uploaded to hackage and is > appropriately > > Trustworthy where needed. > > > > Please let me know if I missed flagged anything you need flagged, or > > mis-flagged anything you think shouldn't be. ;) > > > > > > On Mon, Oct 29, 2012 at 5:42 PM, Michael Sloan <mgsl...@gmail.com> > wrote: > >> > >> It happened somewhere between 2.6 and 2.7: > >> > >> > >> > http://hackage.haskell.org/packages/archive/lens/2.6.1/doc/html/Control-Lens-Internal.html > >> > >> > http://hackage.haskell.org/packages/archive/lens/2.7.0.1/doc/html/Control-Lens-Internal.html > >> > >> The strange thing is that the only internal dependency of 2.7.0.1, > >> Control.Lens.Isomorphic, is still "Safe-Infered" (this spelling error > >> should probably be fixed) > >> > >> > >> > http://hackage.haskell.org/packages/archive/lens/2.7.0.1/doc/html/Control-Lens-Isomorphic.html > >> > >> It'd be pretty fancy if Safe Haskell could give reasons for "None" / > >> tell what would have made it unsafe in the event of "Trustworthy". > >> Particularly fancy if integrated into the haddocks. > >> > >> -mgsloan > >> > >> On Mon, Oct 29, 2012 at 2:33 PM, Edward Kmett <ekm...@gmail.com> wrote: > >> > Actually Control.Lens.Getter doesn't use TH. The issue is more that it > >> > depends on some modules I didn't flag as Trustworthy and which require > >> > some > >> > more high-falutin type system extensions that GHC isn't happy about > >> > treating > >> > as Safe. I'll try adding a few Trustworthy flags. > >> > > >> > It previously was treated as Trustworthy or SafeInfered throughout. > >> > > >> > Somewhere along the way I must have toggled on an extension and broken > >> > that > >> > property. > >> > > >> > -Edward > >> > > >> > > >> > On Mon, Oct 29, 2012 at 1:14 PM, Petr P <petr....@gmail.com> wrote: > >> >> > >> >> Hi > >> >> > >> >> I believe the reason is that it uses TemplateHaskell for automatic > >> >> derivation of labels. And TemplateHaskell is of course unsafe, since > >> >> it could convert your code into something entirely different. > >> >> > >> >> Best regards, > >> >> Petr Pudlak > >> >> > >> >> 2012/10/29 Greg Fitzgerald <gari...@gmail.com>: > >> >> > Why are getters from the 'lens' package unsafe? Is there a subset > >> >> > like Data.Label.Pure from 'fclabels' that can be imported safely? > >> >> > > >> >> > > >> >> > $ cat a.hs > >> >> > {-# LANGUAGE Safe #-} > >> >> > > >> >> > import Control.Lens.Getter > >> >> > > >> >> > main = print 123 > >> >> > > >> >> > $ runghc a.hs > >> >> > > >> >> > a.hs:3:1: > >> >> > Control.Lens.Getter: Can't be safely imported! > >> >> > The module itself isn't safe. > >> >> > > >> >> > > >> >> > Thanks, > >> >> > Greg > >> >> > > >> >> > _______________________________________________ > >> >> > Haskell-Cafe mailing list > >> >> > Haskell-Cafe@haskell.org > >> >> > http://www.haskell.org/mailman/listinfo/haskell-cafe > >> >> > >> >> _______________________________________________ > >> >> Haskell-Cafe mailing list > >> >> Haskell-Cafe@haskell.org > >> >> http://www.haskell.org/mailman/listinfo/haskell-cafe > >> > > >> > > >> > > >> > _______________________________________________ > >> > Haskell-Cafe mailing list > >> > Haskell-Cafe@haskell.org > >> > http://www.haskell.org/mailman/listinfo/haskell-cafe > >> > > > > > >
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
