| Age | Commit message (Collapse) | Author |
|---|
|
Avoids re-updating too frequently -- it's rare that we need a higher
resolution of updates than this, in which case a manual `cabal update`
should suffice.
(Otherwise, if hackage updated between builds, this will trigger another
update)
|
|
In adapting my stack-based PKGBUILDs, forgot cabal doesn't automatically
pull in GHC -- h/t to @andykluger who reminded me this was necessary
|
|
|
|
|
|
|
|
Separate building the package and its deps into build() and prepare(),
respectively (perhaps counterindicated and a sign we should actually be
using the wider Arch Haskell ecosystem? Will revert if I see pushback)
Uses the --offline flag added in cabal 3.12.1.0 -- the flag was added
way back in 1.24.0.0 in 2016-03, so it won't complain on older cabals,
but the functionality for build was only added in 3.12.1.0
|
|
.gitattributes are useful for local configuration, but they shouldn't
have been checked in
|
|
Plus add prepare() for when building in a chroot
|
|
Thanks to @alerque for reminding me of the makedep requirement
|
|
Don't retag for a version if a tag already exists (picked up on by this
causing unexpected output)
|
|
Brittle, and upstream has merged anyway
|
|
* Use patch instead of git am -- latter has some heaviness (but not
enough for its own patch)
NOTE: This is currently commented out since the change has the side
effect of downgrading the package -- git am creates a new commit, and
so it is counted for determining the version, whereas patch just
modifies the working tree.
* Clarify why the weird pkgver()
* Use v2- over new- commands for stability
* No need for $srcdir prefix everywhere
|
|
Previous config also killed diffs
|
|
Accepts PR #32 from upstream
|
|
|