owez / jingo Goto Github PK

Forgot to change after thinking of the name "jingo"

Could be useful (but a little confusing) if docs where made for the jingo/ CLI crate as it will never be publically available on crates.io

Could use with more examples in examples/ to show off more of the general language

Could be faster by just returning a token and appending it inside of the main token scan. This will also allow scan_next_token to be a public function (would need to be renamed to scan_token though).

It is a high level library but contains 8 dependencies, needlessly hurting compile size compared to something lower-level but that works just as well.

Mainly a binary size issue to reduce ideally ~150kb from the binary, potentially halving it's size.

User cannot currently pick anything other then normal, cli is exactly the same to user currently even though it's been expanded under the hood much more

Need to parse & strip multiple lines of --- hi\n--- more\n--- docs\nclass Hi; because right now it's [DocStr, DocStr, DocStr, ...], not [DocStr, ...] in parsing

CLI has became cumbersome to work on in jingo/ crate and should be (partially) rewritten, prehaps with help of the in-house climake once that has been rewritten itself.

Should add a \" so you don't accidently escape strings and such and some basic string literal tests

You will continue to develop, also want to understand this knowledge, develop your own programming language

Could use the main regex library or the lower level regex_automata. Looks like regex has technically no dependancies (but optionals are probably enabled) and the other has 1 normal dep and 2 optional ones that may not be used.

Probably best to use the normal regex but still up for debate.

Got a lot of strange borrowing stuff going on that could be simplified with better lookahead instead of chars.next() and stuff all the time, fixing weirdness of get_numlit_data, get_strlit_data and get_comment_data.

No need for a raw parameter on each token added, just give string representations as fmt::Display. Saves a lot of memory and code uglyness.

Need to implement a basic AST, can base off of https://github.com/Owez/zypo-rs/blob/master/zypo-lib/src/parser/ast.rs but needs metadata (e.g. line numbers and column numbers) stemming from the lexer.

Need to figure out what to do for lexing number literals, prehaps put it in the same type of thing as iterators but if it doesnt follow the iterator pattern and can be parsed as a number it becomes a number literal automatically.

Currently implemented lexer is setup in a strange pseudo-oop way as I was directly translating from the crafting interpreters lexing tutorial with a few differences. Should be changed to a simple function with private functions that get passed args directly.

A very basic parser should be set in the next week or two, most likely using lalrpop as i'm already familiar with it and dev time > dependency count.

Functions like -i/--install or -r/--remove will most likely be used in the future to install/remove packages with the yet un-made package manager. Not yet sure about any futher details about dependencies or if a package manager will be integrated into the binary but we'll see.