This adds support for the extends keyword which lets the tests pass on Build 4075 as requested in #323. The implementation is very hacky right now:
We now require an optional base_syntax when parsing a syntax, if the base_syntax is not present and we find an extends keyword at the top level, we error out reporting which syntax file we expected.
SyntaxSetBuilder handles this error and resolves the syntaxes in a loop until no more syntaxes can be resolved.
We don't handle multiple-inheritance (extends might be a list).
We don't re-evaluate the contexts as defined in the spec (you can override variables and have that affect the context).
We only handle extends for syntaxes added through SyntaxSetBuilder::load_from_folder.
I am opening this PR on top of #535 to get the fixes to the tests.
I don't know if this is a correct implementation of the extends keyword, specifically, I don't know when contexts are evaluated; the contexts are properly merged, but the variables can be overwritten and thus the context needs to be re-evaluated, I don't know if I need to do something about it, or id the context evaluation is done after the SyntaxSet is built.
I am not happy with the implementation either, but I don't know how to improve it further. I will happily improve it if you have any ideas.
Ideally, SyntaxSetBuilder::build would return an error that indicates the syntax was unable to extend another syntax.
This adds support for the
extendskeyword which lets the tests pass on Build 4075 as requested in #323. The implementation is very hacky right now:base_syntaxwhen parsing a syntax, if thebase_syntaxis not present and we find anextendskeyword at the top level, we error out reporting which syntax file we expected.SyntaxSetBuilderhandles this error and resolves the syntaxes in a loop until no more syntaxes can be resolved.extendsmight be a list).contextsas defined in the spec (you can override variables and have that affect the context).extendsfor syntaxes added throughSyntaxSetBuilder::load_from_folder.I am opening this PR on top of #535 to get the fixes to the tests.
I don't know if this is a correct implementation of the
extendskeyword, specifically, I don't know when contexts are evaluated; the contexts are properly merged, but the variables can be overwritten and thus the context needs to be re-evaluated, I don't know if I need to do something about it, or id the context evaluation is done after theSyntaxSetis built.I am not happy with the implementation either, but I don't know how to improve it further. I will happily improve it if you have any ideas.
Ideally,
SyntaxSetBuilder::buildwould return an error that indicates the syntax was unable to extend another syntax.