- Supports Ruby 2.0+ runtime #986
Conversation
This is an alternative approach to the changes in whitequark#985. The Parser gem may be required to operate across as many Ruby versions as possible. The use of `delete_suffix` starts from Parser version 3.3.0.0. Therefore, by replacing `delete_suffix` with `sub`, and releasing Parser 3.3.0.1, supporting Ruby 2.0 would be feasible.
|
Ruby 2.0.0 has reached EOL at |
|
For example, Ruby Next is used in a transpiler that allows for using new Ruby syntax on older Ruby runtimes. It seems to support Ruby 2.2+. There are also users who cannot necessarily move away from EOL versions. Considering that the Parser gem is arguably the most utilized Ruby parser implementation, I think there is worth in supporting older Ruby runtimes as much as possible. On the other hand, I deeply regret that this comes at the cost of not being able to use newer language features 🙇 |
|
Ok, fair enough |
|
I've released |
|
Thank you for considering this matter. |
…2.3 (rmagick#1422)" The parser gem has still support Ruby 2.0+ whitequark/parser#986 This reverts commit 6f906a8.
…2.3 (#1422)" (#1425) The parser gem has still supported Ruby 2.0+ whitequark/parser#986 This reverts commit 6f906a8.
|
It's not clear to me why someone with a very old Ruby would want to use later versions of Nevertheless, if this comes up from time to time, consider using |
I've found whitequark#986 regressions in the following RuboCop CI matrix: https://github.com/rubocop/rubocop/actions/runs/7437077900/job/20234293932?pr=12598 So, whitequark#986 was not a sufficient fix. This PR fixes an error when using heredoc with non-word delimiters: ```ruby <<~'+' foo + ``` The following repro test have been newly added: ```console $ bundle exec ruby -Itest test/test_lexer.rb (snip) 1) Error: TestLexer#test_heredoc_plus_character: RegexpError: target of repeat operator is not specified: /+\z/ /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer/literal.rb:250:in `delimiter?' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer/literal.rb:138:in `nest_and_try_closing' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer-strings.rb:4165:in `advance' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer-F1.rb:11369:in `advance' test/test_lexer.rb:81:in `assert_scanned' test/test_lexer.rb:1001:in `test_heredoc_plus_character' ```
I've found whitequark#986 regression in the following RuboCop CI matrix: https://github.com/rubocop/rubocop/actions/runs/7437077900/job/20234293932?pr=12598 So, whitequark#986 was not a sufficient fix. This PR fixes an error when using heredoc with non-word delimiters: ```ruby <<~'+' foo + ``` The following repro test have been newly added: ```console $ bundle exec ruby -Itest test/test_lexer.rb (snip) 1) Error: TestLexer#test_heredoc_plus_character: RegexpError: target of repeat operator is not specified: /+\z/ /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer/literal.rb:250:in `delimiter?' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer/literal.rb:138:in `nest_and_try_closing' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer-strings.rb:4165:in `advance' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer-F1.rb:11369:in `advance' test/test_lexer.rb:81:in `assert_scanned' test/test_lexer.rb:1001:in `test_heredoc_plus_character' ```
I've found #986 regression in the following RuboCop CI matrix: https://github.com/rubocop/rubocop/actions/runs/7437077900/job/20234293932?pr=12598 So, #986 was not a sufficient fix. This PR fixes an error when using heredoc with non-word delimiters: ```ruby <<~'+' foo + ``` The following repro test have been newly added: ```console $ bundle exec ruby -Itest test/test_lexer.rb (snip) 1) Error: TestLexer#test_heredoc_plus_character: RegexpError: target of repeat operator is not specified: /+\z/ /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer/literal.rb:250:in `delimiter?' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer/literal.rb:138:in `nest_and_try_closing' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer-strings.rb:4165:in `advance' /Users/koic/src/github.com/whitequark/parser/lib/parser/lexer-F1.rb:11369:in `advance' test/test_lexer.rb:81:in `assert_scanned' test/test_lexer.rb:1001:in `test_heredoc_plus_character' ```
these users can also do core extensions to do a "single method" backport if needed. I personally would not regress my own project for these users. |
This is an alternative approach to the changes in #985. The Parser gem may be required to operate across as many Ruby versions as possible. The use of
delete_suffixstarts from Parser version 3.3.0.0. Therefore, by replacingdelete_suffixwithsub, and releasing Parser 3.3.0.1, supporting Ruby 2.0 would be feasible.