* "Misc" use was disputed, moving back to use of the "Additional" category.
* Using a (link) pattern for URLs, reasons being:
* Including the markup into the outlines or hyperlinks considered a
bad practice and should be used carefully.
* The human cause:
The reason for the changelog is so people read through it and get the info.
Hyperlinks make text colored. Studies of readability have shown that
people perceive the information of a brightly colored text
much more poorly than the text of regular coloring.
And delivering the meaning is what changelog text is for, so the
informative text should be regularly colored.
* The technical cause -
the Hackage does not parse the markup inside the hyperlinks:
https://hackage.haskell.org/package/hnix-0.10.0/changelog
Probably because of the reason *1, but Hackage reason alone is enough.
* Since people perceive and think about the read information mostly linearly
and in the FIFO manner, `(link)` is put before information text, so links
are placed in the same space and look more uniform upon reading. And upon
reading the "link" context gets pushed first from the FIFO mind buffer and
the further understanding the text the person is not distracted by
sudden reading (taking-in) of the colored word "link" in the end of the
semantically challenging text that is required to be understood.
* Dependency requirements of last major versions made literal.
* Link to `(diff)` made more obvious. While professionals tend to look for the
link to the full diff, diff under version number is ambiguous to where the
version number link leads, it can be for example: a GitHub release (where
different forms of packages are provided), diff, commit listing (as it was before
the current change), official forum post, official news... etc.
* Made `(diff)` link to point directly towards the diff.
Option for debugging and bugreporting purposes.
Current output:
```
Version: 0.9.1
Commit: 2dc211314e
date: Sat Sep 12 13:31:59 2020 +0300
branch: 2020-09-12-add-GitRev-to-version
```
Also date and branch is to direct contributors attention to updating (or having
in mind) that.
M hnix.cabal
M src/Nix/Options/Parser.hs
* Prepare release 0.9.2
* CHANGELOG: m upd to 0.9.{2,1}
* Update changelog and version
* CHANGELOG: m upd: unification of structure, forming 0.10.1
Breaking changes into "Breaking" section.
Miscellaneous changes into "Misc" section.
Co-authored-by: Simon Jakobi <simon.jakobi@gmail.com>
1.5.3 introduced instances for `strict` & `data-fix`, some intstances overlap.
Trying to remove/sync instances results in:
```text
src/Nix/Expr/Types.hs:1:1: error:
Exception when trying to run compile-time code:
Constructor ‘DynamicKey‘ must only use its last 1 type variable(s) within
the last 1 argument(s) of a data type
```
More files than necessary were included from the nix submodule.
This also adds some missing eval-compare test files. The testsuite
now always runs the same number of tests, no matter the
`buildFromSdist` setting.
This also sorts the files from the submodule, so they are easier to
navigate.
In addition to `interpolate`, this also removes the following
transitive dependencies:
- `haskell-src-exts`
- `haskell-src-meta`
- `safe`
- `th-expand-syns`
- `th-lift`
- `th-lift-instances`
- `th-orphans`
- `th-reify-many`
Since `neat-interpolation`'s `text` quasiquoter trims more whitespace
than `interpolate`'s `i`, the expected output of some tests for
`unsafeGetAttrPos` had to be updated.
This also improves the test failure output of the `constantEqual`
helper.
Fixes#634.
Set current upper bound of lenses
`template-haskell` bounds derived from the git commit date and current bound,
then lowered low bound to include GHC 8.4
M hnix.cabal
M hnix.cabal
The `MonadException` class was removed in favor of `MonadCatch`, so we
can remove its instances, enabling us to remove the library dependency
on haskeline.
To simplify dependency resolution with GHC < 8.10, this also disables
the executable in that configuration.
The `Refl` constructor had previously been re-exported from
`Data.GADT.Compare`.
This also replaces the dependency on `dependent-sum` with `some`,
where the `Data.GADT.Compare` module is now defined.
Tested by building locally with GHC 8.4.4, 8.6.5 and 8.8.3 in `cabal`.
Fixes#585.
"breaking change in 0.6.2.1/0.6.2.2 and properly do major version bump to
reflect the breaking change"
Changelog: https://hackage.haskell.org/package/dependent-sum-0.7.1.0/changelog
0.7.1.0 - 2020-03-25
Shift version bounds for some to 1.0.1.* versions.
0.7.0.0 - 2020-03-24
Fix ChangeLog to include the breaking change in 0.6.2.1/0.6.2.2 and properly do major version bump to reflect the breaking change.
0.6.2.2 - 2020-03-23
Update GitHub repository in cabal metadata.
0.6.2.1 - 2020-03-21
(Breaking change) Removed modules Data.GADT.Compare, Data.GADT.Show, Data.Some and now re-export them from the some package. This forced some deprecations to be fully realized.
Update cabal meta-information (tested with GHC 8.8).
"This forced some deprecations to be fully realized." - it compiles, so
those functions was not used in HNix.
M hnix.cabal
The only actual breaking change affecting `hnix` is `these` splitting
into three packages and renaming some things. Otherwise, as the
package should work fine with either the newer or older versions, I
have left the previous bounds in place and added the new ones
disjunctively.
This will help somewhat with #494.