Closed bergey closed 8 years ago
Yeah, I'm fine with dropping 7.6, especially if we wait until after GHC 8.0 lands to cut a major new release (which seems basically guaranteed at this point).
I wonder if we should start being explicit about which stack resolvers we support.
I notice you didn't remove the line listing GHC 7.4, I assume we should remove that as well?
Yes, we should certainly remove the line for GHC 7.4. It's in allow_failures, so it doesn't bug me as much =).
I'd like to test on Stackage LTS-3; not sure beyond that. That should fill the hole where we don't have HP-2015. I haven't yet tried to make travis run some tests with stack & some with cabal. It shouldn't be very hard. I'll have a PR soon about testing -builder
with stack, since it hasn't passed CI in ages.
Fine by me
On Wed, Dec 2, 2015 at 2:26 PM Daniel Bergey notifications@github.com wrote:
Merged #59 https://github.com/diagrams/diagrams-contrib/pull/59.
— Reply to this email directly or view it on GitHub https://github.com/diagrams/diagrams-contrib/pull/59#event-480585911.
Building on 7.6 has been impossible since 71f6539c300c55dc7f7439bcb582d32c0b54fcdc which added a dependency on cubicbezier, which requires containers >= 0.5.3, whereas GHC-7.6 has template-haskell built against containers 0.5.0.
Are folks OK dropping 7.6 now?
@kuribas - Is it possible for
cubicbezier
to supportcontainers-0.5.0
? How hard would it be?Tangentially, we should probably restrict our travis builds to
template-haskell installed
. The failed builds here try to rebuildtemplate-haskell
with newercontainers
, which I doubt is ever an acceptable solution.