Revision history for Happy

2.1

• Added --numeric-version CLI flag.
• Documented and implemented the new feature "Resumptive parsing with catch"
• Documented (and reimplemented) the "Reporting expected tokens" feature

2.0.2

The 2.0.1 release in turn exposed two more regressions:

• Generated code uses PatternGuards without declaring it (#309)
• Use of happy-lib:* syntax to depend on private library components triggered a bug in Cabal versions 3.0 and 3.2 (#311)

This release fixes both.

2.0.1

The 2.0 release changed the indentation character from tabs to two spaces, triggering an unforced breaking change in GHC (#303). This release provides the fix by using eight spaces for indentation.

2.0

There are two main breaking changes in this release:

1. Removed non-array, non-GHC modes, so flags -ag are the default now and become no-ops.
2. Generated parsers now activate the language extension -XNoStrictData without which every use of a happy parser would lead to an immediate crash (#273). This causes us to drop support for GHC < 8.0.

Furthermore, the project structure was modularized and a library happy-lib containing the implmentation of the happy executable was extracted.

Quite similar to the situation with GHC vs. the GHC API, we expect that happy will continue to be a stable CLI tool with solid (if occasionally out of date) documentation, while the design, documentation and implementation of happy-lib is still in flux and use is only recommended to expert users.

Other, more minor changes:

• Revert the new bootstrapping system of 1.21.0 to mitigate build issues (#255, #274).

• Encode action table offsets in 32 bit instead of 16 bit (#93, #199, #266). This increases the size of generated parsers a bit (about 250KB for GHC's parser), but also manages to generate parsers for grammars that were previously running into the size limit (#199).

• The documentation has been converted to ReStructuredText, hosted at https://haskell-happy.readthedocs.io/en/latest/ (#226)

• A few internal refactorings to the structure of generated code.

1.21.0

The main focus of this release was revamping the build system and bootstrapping.

• The release no longer contains generated source code. Instead of simply requiring a pre-built bootstrap version of Happy in that event, we have a parser-combination-based implementation of enough of Happy to bootstrap the rest. (Currently, the bootstrap version is everything but attribute grammars, and thus sufficient for e.g. GHC, but this is subject to change.) The bootstrap version of Happy is then sufficient to build Happy once again with all features enabled.

  Note, this means users of attribute grammars will have to modify the way they build happy if they were previously building from Hackage relying on the pre-generated sources.

• Rather than creating many "templates" at build time, there is a single combined template. Different implementations are chosen using CPP, as was already done within the templates before.

• Some imports were tightened down, which may help building with newer versions of base.

1.20.1

• Fix for building with mtl-2.3.1 (GHC 9.6)

1.20.0

• Fix #121: the -i flag produces an .info file even if the %expect pragma is violated
• Fix #131: qualify uses of Prelude functions in generated code
• Fix #161: drop fewer parse items when generating an .info file
• Introduce the %shift directive to resolve shift/reduce conflicts explicitly, useful in conjunction with %expect 0
• Remove the deprecated build configuration flag small_base

1.19.12

• Fix for building with GHC 8.8.x
• Move custom Setup preprocessing steps into a separate executable, like Alex

1.19.11

• Fix for building with GHC 8.6.x

1.19.10

• Fix polymorphic (rank-n) non-terminals
• Fix for GHC 8.8.1

1.19.9

• Fix cabal warnings
• Bump upper bounds
• Fix build with GHC 8.4.1-alpha

1.19.8

• Fix issue #94 (some grammars don't compile due to new type signatures introduced to allow overloading to be used)

1.19.7

• Fix missing test suite files in the sdist

1.19.6

• Manually generate Parser.hs using Makefile before sdist, to fix bootstrapping problems with cabal sandboxes & new-build
• Documentation fixes
• Fixed GLR support
• New option -p/--pretty prints the grammar rules (only) to a file
• Added generation of additional type signatures to enable use of typeclasses in monadic parsers.

1.19.5

• Fixes for GHC 7.10
• Code cleanups (thanks Index Int vlad.z.4096@gmail.com)

1.19.4

• Fix for GHC 7.10 (Applicative/Monad, #19, #21)

1.19.3

• Fix for GHC 7.2 (#16)

1.19.2

• Fixes for clang (XCode 5)

1.19.1

• Repackaged to build with GHC 7.7+

1.19

• Necessary changes to work with GHC 7.8

1.18.10

• Fix build with GHC 7.6

1.18.8

• Fix a packaging bug (cabal-install-0.10.2 didn't put the Happy-generated files in the sdist)

1.18.7

• Fix a bug in error handling when using %monad without %lexer

1.18.5 --- 17 Jun 2010

1.18.4 --- 23 April 2009

1.18.2 --- 5 November 2008

1.18.1 --- 14 October 2008

1.18 --- 13 October 2008

• New feature: EBNF-style paramterized macros, thanks to Iavor Diatchki.
• Works with Cabal 1.2, 1.4 and 1.6
• A few minor bugfixes

1.17 --- 22 October 2007

• Cabal 1.2 is required
• Works with upcoming GHC 6.8.1
• Fix the parE bug (poor error message for errors in the grammar)
• Some performance improvements to Happy itself

1.16 --- 8 January 2007

• Switch to a Cabal build system: you need a recent version of Cabal (1.1.6 or later). If you have GHC 6.4.2, then you need to upgrade Cabal before building Happy. GHC 6.6 is fine.

• New %error directive

• New production forms: {%% .. } and {%^ .. }

• Added Attribute Grammar support, by Robert Dockins

1.15 --- 14 January 2005

• New %expect directive
• The list of tokens passed to happyError now includes the current token (not %lexer).
• Added support for ambiguous grammars via Generalized LR parsing
• Added %partial to indicate a parser that can return a result before EOF is reached.

1.14 --- 14 April 2004

• New meta-variable $> represents the rightmost token.

• Happy's OPTIONS pragma is merged with an existing one in the grammar file, if any.

1.13 --- 19 June 2002

• Support for newer versions of GHC (>= 5.04).

• Addition of an experimental flag: --strict.

1.11 --- 25 September 2001

• Tokens no longer have a default precedence --- if you want a token to have a precedence, you have to declare it.

• Bugfix to templates for GHC on 64-bit platforms.

1.10

• Bugfixes, and minor performance improvements,

• Most of the examples work again.

1.9

• A grammar may now contain several entry points, allowing several parsers to share parts of the grammar.

• Some bugfixes.

1.8

• Parser table compression, and more efficient table encoding when used with GHC. Large grammars can now be compiled in much less time/space than before using GHC.

• Yacc-style operator precedence, thanks to patches from Hermann Oliveira Rodrigues hermann@dcc.ufmg.br and Josef Svenningsson josefs@cs.chalmers.se.

• A debug option which causes the generated parser to print tracing information at each step during parsing.

1.6

• Now written in, and generates, Haskell 98.

• Several bug fixes.

• A new option, -c, generates parsers that use GHC's unsafeCoerce# primitive to speed up parsing and cut down the binary size. The -c option can only be used with the -g (GHC extensions) option.

• Parsers generated with the -g option will compile to smaller binaries now --- some sources of parser-bloat were identified and squished.

• Happy has a new Open Source license, based on the BSD license.

• A sample Haskell parser using Happy is included.

1.5

• Many bug fixes to the error recovery support, found by experimenting with the Haskell grammar and layout.

• Happy is about 5 times faster on large examples, due to some changes in the LALR(1) algorithms.

As of version 1.5, Happy is capable of parsing full Haskell. We have a Haskell parser that uses Happy, which will shortly be part of the library collection distributed with GHC.

1.2

• Supports Haskell 1.4

• Lots of bugs fixed

• Performance: the parser generator is at least 20% faster, and generated parsers should be faster due to the replacement of a data type with a newtype.

• Simple error recovery: designed to be enough to implement the Haskell layout rule.

• Revamped monad support: the monad can now be threaded through the lexer, enabling passing of state between the parser and the lexer (handy for the Haskell layout rule), and consistent error handling.

• The %newline feature is removed, the same effect can be achieved using the new monad support.

0.9

• Happy should be much faster than before.

• Generated parsers will be 5-10% smaller.

• Happy now compiles with ghc-0.26.

• Support for monadic parsers via %monad (see the documentation).

• New syntax: previously

  f :: { <type> }
  f : ...
    | ...
    etc.
  

  can now be written

  f :: { <type> }
    : ...
    | ...
    etc.
  

  (i.e. omit the extra f. It was always ignored anyway :-)

• Miscellaneous bug fixes.