Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
* Supporting different field names for tags.
* Removed buildTagsByField
Removed to avoid exponential growth of helper functions.
Renamed field parameter in getTagsByField to fieldName
to avoid shadowing.
* Drop obsolete export
Co-authored-by: Alexander Batischev <eual.jp@gmail.com>
|
|
Report and fix by @alexandroid000. Fixes #860.
|
|
* update Github Actions
https://github.com/actions/setup-haskell has been archived.
I believe https://github.com/haskell/actions/tree/main/setup is current.
* try to fix stalling CI job
|
|
Compilers that use data from sources other than local files may need
to be recompiled, but Hakyll's file-based dependency checking does
not handle this situation.
Add a new kind of dependency called 'AlwaysOutOfDate'. If an item
has this dependency, it will be unconditionally rebuilt.
Also add the 'forceCompile' rule modifier, which is a user-friendly
way to force recompilation of specific items. Example usage:
forceCompile $ create ["foo"] $ do
route $ idRoute
compile $ unsafeCompiler $ doSomeIO
|
|
|
|
* Allow file-embed 0.0.14.0
* Allow time 1.11
* Allow template-haskell 2.17 (i.e. GHC 9)
|
|
* Bump allowed Pandoc version
* Allow more recent version of Pandoc in hakyll-init exe
|
|
* Async runtime
* Activate multi-threading in template repo
* Style changes after feedback
* Limiting the number of concurrent tasks
* Revert "Limiting the number of concurrent tasks"
This reverts commit 38984f6f5332632be8c4cab3e29d37e318492d70.
|
|
Fixes #847.
|
|
There's no mention of this on the issue tracker, so here's my recap of
what led to this change.
There is some drama around Freenode. Their volunteer staff quit[1]. Then
the new-ish management started enacting strange policy changes[2] and
take over the channels[3]. On June 26th 2021 Freenode's bot tried to
take over #hakyll, but failed; however, it did succeed in #haskell and
many, many other channels.
For the preceding week, me and IRC user henk were trying to move the
channel off the Freenode, either to Libera.Chat or OFTC. Jasper, the
founder of Hakyll, was absent from IRC and didn't respond to my emails.
Me and henk are the most active IRC users on the channel. Also, I have
Collaborator access to the repo, and henk has chanop access to the IRC
channel. We believe that at this time, we're the ones who are the best
positioned to execute the move, so we're doing it.
We only considered Libera.Chat and OFTC. Libera.Chat was created a week
earlier by the same staff who quit Freenode; I personally consider them
to be a spiritual successor of Freenode. OFTC is a well-established
network with good governance documentation. Both networks are
FOSS-friendly. The choice wasn't obvious.
Libera.Chat was picked because 10 users moved there (and 1 more did
while I was typing this!), whereas only 2 joined OFTC (me and henk, and
those weren't votes for OFTC — we were just ensuring that the channel is
ours). Also, #haskell and #ghc are on Libera too, so it makes sense for
a Haskell project such as Hakyll to be present on Libera.
For posterity:
<Minoru> info #hakyll
-ChanServ(ChanServ@services.)- Information on #hakyll:
-ChanServ(ChanServ@services.)- Founder : jaspervdj
-ChanServ(ChanServ@services.)- Registered : Mar 01 13:29:17 2011 (10y 12w 5d ago)
-ChanServ(ChanServ@services.)- Mode lock : +ntc-slk
-ChanServ(ChanServ@services.)- *** End of Info ***
1. https://kline.sh
2. https://github.com/freenode/web-7.0/pull/513
3. https://www.devever.net/~hl/freenode_abuse
|
|
|
|
|
|
This passed on my machine:
cabal test all -j --enable-tests --constrain 'pandoc == 2.12'
|
|
All tests are passing here with the new versions.
|
|
|
|
|
|
sites, this can be a major speedup (see #786) (#810)"
This reverts commit 84157674d955778c806efdafda311b2732242b38.
|
|
* added nliu.net to examples for shameless self-promotion
* fixed comma formatting
|
|
* added function that allows for pre-pandoc transformations
* modified haddock docstring
* refactoring to avoid importing the kleisli fish
* Renamed `applyPandoc` to `renderPandocWithTransformM`, added non-monadic
variant. Exported previous new functions that I forgot to export.
|
|
|
|
This reverts commit 87e93c6c95b86b435b532286e07b0a9b896aef8f.
I screwed up with that one: the change in behaviour was a regression[1]
in Pandoc, and I shouldn't have papered over it. It's fairly unlikely
that someone would build Hakyll's test suite with one of the two Pandoc
versions that regressed, so I simply revert my earlier commit.
1. https://github.com/jgm/pandoc/issues/6966
|
|
The new version started wrapping Biblio references into <p> tags, so
we now keep two different golden files for different Pandoc versions.
I had to add Pandoc to dependencies of the test suite, otherwise GHC
won't define `MIN_VERSION_pandoc` macro.
|
|
* Add documentation for pandoc
* Fix grammar
|
|
|
|
* Pandoc 2.11 compatibility
* Bump stack.yaml
* Bump stack dependencies
|
|
|
|
|
|
This removes the need for a workaround on Windows.
|
|
|
|
|
|
|
|
can be a major speedup (see #786) (#810)
Co-authored-by: gwern <gwern@gwern.net>
|
|
|
|
* Bump pandoc version
* Bump Pandoc version for hakyll-website
Co-authored-by: Alexander Batischev <eual.jp@gmail.com>
|
|
|
|
stable release (#798)
|
|
Builds fine and all tests pass here.
|
|
Builds fine and all tests pass here.
|
|
|
|
* Bumped bound for pandoc and add Binary-instances for new
constructors that were added to pandoc-types
* Support most recent template-haskell release
* Set lower bound of pandoc to version 2.10
* Update CI configuration
* Bump QuickCheck upper bound to 2.15
Co-authored-by: OC4 <oc4@protonmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|