2019-06-24 17:06:27 +03:00
|
|
|
Contributing to Idris 2
|
|
|
|
=======================
|
|
|
|
|
|
|
|
Contributions are welcome! The most important things needed at this stage,
|
|
|
|
beyond work on the language core, are (in no particular order):
|
|
|
|
|
2019-06-24 17:16:49 +03:00
|
|
|
* CI integration.
|
|
|
|
* Documentation string support (at the REPL and IDE mode).
|
2019-07-05 18:40:08 +03:00
|
|
|
* Generating HTML documentation from documentation strings (and possibly other
|
|
|
|
formations)
|
2019-06-24 17:06:27 +03:00
|
|
|
* A better REPL, including:
|
2019-06-30 17:50:58 +03:00
|
|
|
- `it` and `:let`
|
2019-06-24 17:06:27 +03:00
|
|
|
- readline and tab completion
|
2019-07-01 01:54:50 +03:00
|
|
|
- `:search` and `:apropos`
|
2019-06-24 17:06:27 +03:00
|
|
|
- help commands
|
2019-07-09 00:48:39 +03:00
|
|
|
- code colouring
|
2019-08-06 16:54:27 +03:00
|
|
|
- it'd be nice if Ctrl-C didn't quit the whole system!
|
2019-07-09 00:48:39 +03:00
|
|
|
* IDE mode improvements
|
|
|
|
- Syntax highlighting support for output
|
|
|
|
- Several functions from the version 1 IDE protocol are not yet implemented,
|
|
|
|
and I haven't confirmed it works in everything.
|
2019-07-14 13:23:58 +03:00
|
|
|
- (Not really part of Idris 2, but an editing mode for vim would be lovely!)
|
2019-07-01 01:54:50 +03:00
|
|
|
* Some parts of the Idris 1 Prelude are not yet implemented and should be
|
|
|
|
added to base/
|
2019-06-24 17:16:49 +03:00
|
|
|
* Partial evaluation, especially for specialisation of interface
|
|
|
|
implementations.
|
2019-07-01 01:54:50 +03:00
|
|
|
* The lexer and parser are quite slow, new and faster versions with better
|
|
|
|
errors would be good.
|
2019-07-09 00:48:39 +03:00
|
|
|
- In particular, large sections commented out with {- -} can take a while
|
|
|
|
to lex!
|
2019-06-24 17:16:49 +03:00
|
|
|
* An alternative, high performance, back end. OCaml seems worth a try.
|
2019-06-25 23:27:46 +03:00
|
|
|
* JS and Node back ends would be nice.
|
2019-06-24 17:06:27 +03:00
|
|
|
|
2019-07-08 19:10:47 +03:00
|
|
|
The default Prelude describes the rationale for what gets included and what
|
|
|
|
doesn't. Mostly what is there is copied from Idris 1, but it's not impossible
|
|
|
|
I've made some silly mistakes on the way. If you find any, please let me know
|
|
|
|
(or even better, submit a PR and a test case!).
|
|
|
|
|
2019-06-24 17:06:27 +03:00
|
|
|
Some language extensions from Idris 1 have not yet been implemented. Most
|
|
|
|
notably:
|
|
|
|
|
|
|
|
* Type providers
|
|
|
|
- Perhaps consider safety - do we allow arbitrary IO operations, or is
|
|
|
|
there a way to restrict them so that they can't, for example, delete
|
|
|
|
files or run executables.
|
|
|
|
* Elaborator reflection
|
|
|
|
- This will necessarily be slightly different from Idris 1 since the
|
|
|
|
elaborator works differently. It would also be nice to extend it with
|
|
|
|
libraries and perhaps syntax for deriving implementations of interfaces.
|
2019-07-18 17:32:47 +03:00
|
|
|
* `%default` directives, since coverage and totality checking have not been well
|
|
|
|
tested yet.
|
2019-06-24 17:06:27 +03:00
|
|
|
|
|
|
|
Other contributions are also welcome, but I (@edwinb) will need to be
|
|
|
|
confident that I'll be able to maintain them!
|
|
|
|
|
2019-07-01 01:54:50 +03:00
|
|
|
If you're editing the core system, or adding any features, please keep an
|
|
|
|
eye on performance. In particular, check that the libraries build and tests
|
|
|
|
run in approximately the same amount of time before and after the change.
|
|
|
|
(Although running faster is fine as long as everything still works :))
|
|
|
|
|
2019-07-01 23:35:19 +03:00
|
|
|
Libraries
|
|
|
|
---------
|
|
|
|
|
|
|
|
Further library support would be very welcome, but unless it's adding something
|
|
|
|
that was in `prelude/` or `base/` in Idris 1, please add it initially into
|
|
|
|
`contrib/`. (We'll reorganise things at some point, but it will need some
|
|
|
|
thought and discussion).
|
|
|
|
|
|
|
|
Think about whether definitions should be `export` or `public export`. As
|
|
|
|
a general rule so far, type synonyms and anything where the evaluation
|
|
|
|
behaviour might be useful in a proof (so very small definitions) are
|
|
|
|
`public export` and everything else which is exported is `export`.
|
|
|
|
|
|
|
|
Syntax
|
|
|
|
------
|
|
|
|
|
2019-06-24 17:06:27 +03:00
|
|
|
Some syntax that hasn't yet been implemented but will be:
|
|
|
|
|
|
|
|
* Idiom brackets
|
|
|
|
* !-notation [needs some thought about the exact rules]
|
|
|
|
|
|
|
|
Some things from Idris 1 definitely won't be implemented:
|
|
|
|
|
2019-07-18 17:32:47 +03:00
|
|
|
* `%access` directives, because it's too easy to export things by accident
|
2019-06-24 17:06:27 +03:00
|
|
|
* Uniqueness types (instead, Idris 2 is based on QTT and supports linearity)
|
|
|
|
* Some esoteric syntax experiments, such as match applications
|
2019-06-25 23:27:46 +03:00
|
|
|
* Syntax extensions (at least in the unrestricted form from Idris 1)
|
|
|
|
* DSL blocks (probably) unless someone has a compelling use case that can't
|
|
|
|
be done another way
|