chore: Release lalrpop-util version 0.23.1

Bugfixes

• Fix bug where windows paths were used based on target OS rather than host OS, breaking cross compile scenarios

<a name="0.23.0"></a>

chore: Release lalrpop-util version 0.23.0

Breaking Changes

• Lalrpop no longer exposes a regex-automata feature. This feature was previously exposed accidentally and did not change lalrpop functionality.
• Make the lexer feature a default feature for lalrpop-util. In older versions of rust, lalrpop-util would automatically use the lexer feature when lalrpop did, but that is no longer the case, so this change keeps the common case as the default. If you're using a custom lexer, you'll need to make sure to disable default-features for both lalrpop and lalrpop-util.
• Configuration.set_in_dir() can now no longer affect the output dir. In certain configurations, setting in_dir using set_in_dir() could actually cause the generated parser to be put in the input directory rather than in OUT_DIR. In most cases OUT_DIR may be what you want anyways and should work seamlessly. If you want to preserve the existing behavior of writing to the source directory, you should do that explicitly with set_out_dir() (or use_cargo_dir_conventions())
• Configuration.set_in_dir() is now incompatible with process_current_dir(), and process_dir() , because all of these functions set an input directory themselves (either the current directory, or the argument). Previously, these functions would all silently ignore the set_in_dir() setting, except possibly to change the output directory as mentioned above. Now these functions are incompatible. If you get an error about this, remove set_in_dir() from your Configuration, or use it with process(). You may also need to change your output directory as mentioned in the previous bullet.

Features

• lalrpop_util::ParseError now implements the Hash trait
• Support expanding multiple anonymous format arguments in the same pattern (e.g. <A> <B> => format!("<> <>")). It is an error if the number of format strings in the pattern do not match the number in the action code.
• Support binding tuple patterns in grammar. This syntactic sugar can make dealing with nonterminal definitions that return multiple values easier.

Changes

• Bump MSRV to 1.85
• Update to the 2024 rust edition

Internal Development features

• Support running tests with cargo-nextest.
• Support running benchmarks on both lalrpop build time and generated parsers using cargo bench

Bugfixes

• Documentation updates
• When using a custom lexer, the Location type previously required the Copy trait, which was undocumented and unintentional. This requirement has been dropped.
• If looking for a .lalrpop file in a directory that no longer exists, lalrpop no longer creates the missing directory.

<a name="0.22.2"></a>

chore: Release lalrpop-util version 0.22.2

Features

• Support \0 and \x## ASCII escape sequences in grammars
• Documentation updates

Bugfixes

• Avoid clippy warnings for uninlined format strings in generated code

<a name="0.22.1"></a>

When running cargo doc, the include_str command works fine, however,
cargo publish relocates the package into target/package, so the relative
path to the README is different, causing cargo publish to fail.

External discussion at:

rust-lang/cargo#13309
https://users.rust-lang.org/t/include-str-does-not-work-when-releasing-because-of-changed-pathes/15551/14

The solution suggested in those links is to symlink the top level README
inside your package, and then consistently refer to the README in the
package. That README will be copied into the right place during cargo
publish.

Bugfixes

• Many documentation updates.
• Internal cleanups and testsuite enhancements.
• Isolate a few clippy lint allows more tightly to generated code so that these lints can trigger on action code.
• Bump some dependency versions

<a name="0.22.0"></a>

• Release version 0.22.0

• Fix version in documentation comment

• Update changelog based on recent merges

• Update Cargo.lock to the latest version of dependencies

Breaking changes

• The lexer feature no longer implies the std feature. Now lexer is usable in no_std environments. In no_std, ParseError only implements the Error trait in rust 1.81 or later (since core::error was stablized in 1.81).

Features

• Overhaul cfg attributes. You can now include or omit grammar rules and alternatives based on cargo features with not(), any() and all() support

Bugfixes

• Improvements to error message reporting to improve clarity and suppress extra noise
• lalrpop_mod!() now handles imports correctly
• Reenable some warnings on user code for custom lexers

<a name="0.21.0"></a>

chore: Release lalrpop-util version 0.21.0

Since the last release, a fair number of the commits have been focused on cleaning up and improving LALRPOP's documentation. Shout out to Yudai Takada, George White, and Dinu Blanovschi.

Features

• LALRPOP now throws an error in more cases where it would previously just write out an error message and exit.
• lalrpop::process_src is now the recommended function to use in build.rs files. Previously the documentation incorrectly suggested that lalrpop::process_root looked in ./src instead of .

Bugfixes

• A long-standing bug where LALRPOP would throw a "no entry found for key" exception when trying to handle certain grammars has been resolved.
• LALRPOP will stop expanding macros infinitely during build time via a new macro_expansion_limit.

Compatibility note

Adding a limit to the number of times that LALRPOP will attempt to expand a macro is technically a breaking change. However, the default limit of 200 should be more than enough for the grammars we are currently aware of (which almost always need a limit of less than 5). This limit is customizable via Configuration::set_macro_expansion_limit.

If you have a grammar that uses a significant amount of macro expansion steps, we would be very interested in a PR that adds it to the test suite.

<a name="0.20.2"></a>

• Release 0.20.2

• undo autoformatting of comments in Cargo.toml

Co-authored-by: Yann Hamdaoui [email protected]

Co-authored-by: Yann Hamdaoui [email protected]

Special thanks to our newest maintainers, Daniel Burgener and Patrick LaFontaine for helping to coordinate this release.

Features

• Lalrpop no longer depends on the is-terminal crate (thanks to Kmeakin!)
• Better performance with the default lexer using the underlying regex-automata crate (thanks to QuarticCat!)
• Allow the catch-all _ case for token matching can now be set to a higher precedence in match (thanks to fpoli!)
• Fewer clippy lints triggered in generated code
• Lalrpop now traverses symlinks to find .lalrpop files(thanks mbid!)
• Lalrpop now supports block comments including nestings(thanks seanbright!)

Bugfixes

• Lalrpop now uses the ascii-aware space regex when the unicode feature is not enabled (thanks to QuarticCat!)
• Dangling symlinks in crate no longer cause build failure (thanks to legeana for the report!)
• Unicode is now set as a default feature in lalrpop-util to align with lalrpop's defaults

Compatibility note

• MSRV increased to 1.70.
• process_root_unconditionally now correctly lints as having been deprecated.
• Internal types which lead with a __ and should not be relied upon are no longer publicly exposed (thanks to arnaudgolfouse!)
• Lalrpop files containing a space in their name now return an error.

<a name="0.20.1"></a>

• Release 0.20.1 notes

• Add set -e -o to version.sh

• Update version numbers

• Create bugfix section and msrv nit

• Add new maintainer note... which copies the 0.19.9 maintainer note

Yanked

<a name="0.20.0"></a>

Bringing back 0.19.10 patches and further enhancement.

Breaking changes

• Types and enums name with capitalized acronyms are renamed to camelCase
• Minimum rust supported version is now 1.64
• New unicode feature is added to enable regex/unicode and regex-syntax/unicode. Building lalrpop with --no-default-features may be broken without adding it.

Features

• faster compilation time by up to 2x
• expected tokens in failed parses are more accurate
• support for unicode when using builtin tokenizer

Bug fixes and other changes

• fewer warnings about clippy/unused imports in generated code
• updated to edition 2021
• updated mdbook
• Use inclusive ranges for DFA/NFA
• A new document "Lexing raw delimited content"

<a name="0.19.12"></a>

• Add unicode feature to regex-syntax (thanks to Jan Niehusmann!)

Compatibility note

This is actually not fixing a lalrpop bug but fixing user side missing dependency. lalrpop doesn't directly depend on the feature. But regex from user code probably contains it. We will drop this dependency again in the next release.

<a name="0.19.11"></a>

This release is based on 0.19.9, not 0.19.10 No feature added or other bug fixed since 0.19.9

• Fix regex version dependency issues (thanks to Wilfried Chauveau!)

<a name="0.19.10"></a>

Note: This version is yanked. These changes will be included in 0.20

This release fixes an incompatibility with regex.

Features

• faster compilation time by up to 2x
• expected tokens in failed parses are more accurate
• support for unicode when using builtin tokenizer

Bug fixes and other changes

• fix a compatibility issue with regex 1.8
• fewer warnings about clippy/unused imports in generated code
• updated to edition 2021
• updated mdbook

<a name="0.19.9"></a>

This release addresses backwards compatibility warnings and security alerts, along with a few smaller features.

Special thanks to our newest maintainers, Yann Hamdaoui and YunWon Jeong for helping to coordinate this release.

Features

• stream errors to the terminal as they are computed (thanks to Anders Kaseorg!)
• lex raw identifiers (thanks to Karl Meakin!)
• upgrade to the 2018 edition (thanks to blakehawkins!)

Bug Fixes

• fix bug in lane table state splitting (thanks Anders Kaseorg!)
• fix warnings, typos, and links (thanks ggsh, Kian-Meng Ang, Ian Alexander Joiner, absurdhero, and Ömer Sinan Ağacan!)
• fix licensing to be spdx-conformant (thanks chayleaf!)
• fixs to the whitespace example (thanks Thalia Archibald!)

Security patches

• update regex to 1.5.5
• update thread-local to 1.1.7
• update to is-terminal from atty (thanks kpcyrd!)

<a name="0.19.8"></a>

Features

• update petgraph to v0.6.0 (3b482293)

<a name="0.19.7"></a>

<a name="0.19.6"></a>

<a name="0.19.5"></a>

• Fix inlining of fallible productions
• Update dependencies

<a name="0.19.4"></a>

<a name="0.19.4"></a>

• Allow string literals in patterns #557
• Enable precedence annotations #555
• Shake dependency tree #559

<a name="0.19.1"></a>

• Reduce work for LLVMs inliner

<a name="0.19.0"></a>

Performance

• Move the symbol mismatch panic into a colder path (0c69e999)
• Avoid subtracting in goto (8a47ed8c)
• Emit the GOTO table as nested matches (c5070af2)
• parse_table: Avoid generating unused rows in the matrix (688b9193)

Features

• Use FnMut/FnOnce in ParseErrors map functions (8f73c9dc)

Bug Fixes

• Don't include whitespace in the span with empty nonterminals (11a50e70)

<a name="0.18.1"></a>

Bug Fixes

• Remove eprintln which I thought were removed (a9a775eb)

<a name="0.18.0"></a>

Features

• Allow the tokenizer to contain custom skip regexes/literals (ee2f7060)
• states does not need to be passed to reduce actions (c156b4b2)
• action does not need to be passed to reduce actions (f69bce30)
• Only generate simulate_reduce if error recovery is used (d0a3ccba)
• Accept slices as types (#507) (c3e1cda5, closes #493)

Bug Fixes

• Avoid emitting redundant parentheses (b165fc93, closes #493)

Breaking changes

• The lexer feature is now necessary when lalrpop generates the lexer.

<a name="0.17.2"></a>

Performance

• Avoid some memcpying in reductions (4968e5a6)

Bug Fixes

• Allow the deprecated use of trim_left (bdd65184, closes #428)
• Don't make generated files read-only (0c67bbed)

<a name="0.16.2"></a>

Version 0.16.2 (2018-11-22)