Make it compile with GHC 7.8

[daoo]

The mkIntLitInt type change breaks building with GHC < 7.8. I don't know if that is the best way to fix it either. Feedback requested.

[JohnLato]

I think the fix is fine, but unfortunately it'll break backwards compatibility. The easiest way to maintain compatibility is via CPP guards. Sigh... The other option is to fork that file and put the versioning logic in the cabal file. Which is also rather ugly.

kfish, are you interested in maintaining this codebase? IMHO just merge the request as-is, fix Travis, and cut a new release for ghc-7.8.

[JohnLato]

Also I haven't checked ghc-7.10, but I wouldn't be surprised if something else in TH or plugins has changed that would similarly require CPP.

[daoo]

With forking the file you mean like for PrimRules. That fork seems unnecessary though as the files contains the same code?

[JohnLato]

The files wouldn't contain the same code. One version would contain code
for ghc < 7.8, and one version with your submitted change. I agree it's
probably more sensible to use CPP here...

On 01:00, Fri, Feb 27, 2015 daoo notifications@github.com wrote:

With forking the file you mean like for PrimRules. That fork seems
unnecessary though as the files contains the same code?

—
Reply to this email directly or view it on GitHub
#4 (comment)
.

[daoo]

Yes of course. My intent was to make an somewhat related comment about the case for the modules ConstMath.PrimRules.V704 and ConstMath.PrimRules.V706, which are identical except for the name. Technically, they can be merged, but that's for another pull request/bug report.

[FranklinChen]

Released version of this package doesn't compile with GHC 7.10.2 either, of course. (Just found out about this plugin through http://stackoverflow.com/questions/14047241/simple-example-for-impredicativetypes/14065493#14065493 )