All Versions
Latest Version
Avg Release Cycle
41 days
Latest Release

Changelog History
Page 1

  • v0.12.3 Changes

    • Add scss, sass, systemverilog, orgmode.

    • Update xml definitions from upstream: bash, css, python, r, zsh.

  • v0.12.2 Changes

    • Parser: handle context shift to external context, e.g., BashOneLine##Bash. Closes #139 (issue with Dockerfile).

    • Update xml files from upstream: bash, cmake, markdown, objectivecpp, php, sql-postgresql, sql, stan, zshr

    • Fix formatting in Color doc-comment (Janek Spaderna).

    • Add stan.xml (Brian Ward).

  • v0.12.1 Changes

    • Update syntax definitions: bash, cmake, dockerfile, gnuassembler, markdown, spdx-comments.

    • Support hex escapes using \x in regex char classes (#135). These occur in a number of syntax definitions and weren't correctly interpreted before. Thanks to @Agnishom.

    • Support regex property syntax, e.g. \p{Lu}.

    • Regex: support \B (non-word-boundary).

  • v0.12 Changes

    • Properly handle include elements in keyword lists (#124). A number of syntaxes (e.g. typescript, scala) include keyword lists from other syntaxe,s and previously we weren't able to handle this.

      There are several pieces to this change. We need to store lists where other Syntaxes can look them up, so we add an sLists field to Syntax [API change], and modify the parser to fill this. We change lists so that their values are not just a Text, but a ListItem that can either be a textual value or an include directive, specifying a ListName (syntax name and list name).

      The Keyword constructor for Matcher now takes, instead of a WordSet, Either a ListName or a WordSet (API change).

      Skylighting.Parser now exports resolveKeywords (API change), which modifies all Keyword matchers in a syntax so that Left values with a ListName become Right values with resolved WordSets. The tokenizer applies this function automatically to the SyntaxMap given in Config. But it is more efficient to do this conversion just once, rather than every time tokenize is called. So we have loadSyntaxesFromDir call it on the SyntaxMap. With this optimization, there is not an appreciable performance cost to the changes described above.

    • Skylighting.Regex: Fix bug with regexes like a{10}b (#133). This requires exactly 10 a's; previously we interpreted it as "at least 10."

    • skylighting-extract: take a directory as argument rather than files. This allows us to use loadSyntaxesFromDir.

    • Update xml syntax definitions from upstream: julia, cmake, cpp, isocpp, markdown, python, toml.

  • v0.11 Changes

    • Skylighting.Regex: Support regex subroutines (#118). For example, (?1) is replaced by the regex in the first capturing group. So far we only support this simple, absolute form, not the relative form (?-1) supported by some engines (but not used, I think, in KDE's syntax highlighters). This change involves an API change: Regex in Skylighting.Regex has a new Subroutine constructor, and the Recurse constructor has been removed. Instead of Recurse we use Subroutine 0, which unifies the code.

    • Skylighting.Regex: handle e.g. [\1] and [\123] (without initial 0) as octal escapes (#118). These occur in the zsh.xml syntax definition.

    • Pull xml definitions for bash, cmake, python, zsh from upstream.

    • README: Add a note about pulling syntax definitions from upstream (#138). Update build instructions for recent cabal versions (#131).

  • v0.10.5 Changes

    • Fix regression from 0.10.3 with Haskell highlighting of Char (#120).

    • Update xml syntax definitions from upstream. Updated syntaxes: abc apache asp bash bibtex boo c changelog clojure cmake commonlisp cs curry d diff djangotemplate doxygen elixir elm email erlang fortran-fixed fortran-free fsharp hamlet haskell haxe html idris ini javascript-react javascript json julia kotlin latex lex lilypond literate-curry literate-haskell makefile markdown mathematica maxima mediawiki metafont modula-2 mustache nasm nim noweb ocaml octave opencl perl powershell prolog pure python r roff ruby rust sed spdx-comments sql-mysql sql-postgresql tcl tcsh toml typescript verilog xml xslt xul yacc yaml. Not updated: rhtml (causes an error on our test suite), zsh (has a regex we can't parse).

  • v0.10.5.2 Changes

    • Added swift grammar definition (Igor Ranieri).

    • Simplify instructions for two-step build.

    • Fix link to KDE documentation.

  • v0.10.5.1 Changes

    • Regex: Allow lookaheads to capture groups. Previously captures in lookaheads, like (?=(a*)), were ignored. This led to errors highlighting xml and probably other formats (#121).

    • Throw an exception if a capture group isn't defined (with 'dynamic') rather than simply having getCapture fail so that the rule fails. This will make it easier to debug issues like #121.

  • v0.10.4 Changes

    • Move from hxt to xml-conduit for XML parsing.

      This gives about a 4X speedup in parsing syntax definitions. It also reduces the pandoc build dependency footprint, as we depend on xml-conduit anyway and now no longer need to compile HXT and its dependencies.

      There are improvements in accuracy as well: the change to xml-conduit improved parsing for one of the prolog rules; a matcher that should have been for tab characters had been set for a space instead.

    • Removed some unnecessary build-depends in skylighting-extract.

  • v0.10.4.1 Changes

    • Fixed logic for checking line-end-context (#119).

    • Use NonEmpty for the context stack.

    • Remove unneeded build-depends.