These are chat archives for coala/coala-bears

3rd
Nov 2017
John Vandenberg
@jayvdb
Nov 03 2017 07:16
AnnotationBear is a horrible parser that keeps growing ;-(
Rishav Kumar
@aptrishu
Nov 03 2017 09:13
:-/
John Vandenberg
@jayvdb
Nov 03 2017 09:15
it should just use the Python ast, which provides most of that information
Rishav Kumar
@aptrishu
Nov 03 2017 09:19
But that would not suffice, we are exploring pygment's lexer that makes things easy(100 -> 10 lines of code).
John Vandenberg
@jayvdb
Nov 03 2017 09:27
it is good as a general parser, but pitiful for precision, as any RE based parser is
it is a great option for languages we never want to write our own tools for, but for Python 3, no ... we need the real parser, the only parser that matters, the ast.
Sanket Dasgupta
@SanketDG
Nov 03 2017 09:29
How would that work for other languages?
or right now AnnotationBear is python only?
John Vandenberg
@jayvdb
Nov 03 2017 09:40
AnnotationBear is multi-language , and that is where using pygment's lexers will be a good option, if only increasing the number of languages extremely quickly, but like any generalist, it is not good at any of them.
Mischa Krüger
@Makman2
Nov 03 2017 22:49
it seems it evolves into a bunch of custom parsers^^
which is properly easier to handle because you can focus on a single language and it's more modular then^^ one language -> one parser
maintenance overhead increases, but maybe it's worth it, don't know^^
John Vandenberg
@jayvdb
Nov 03 2017 23:25
building custom parsers for individual languages is more work, so we'd only do it for languages that we care a lot about. i.e. Python 3.