Say we have a module that import/open the Prelude:
Test.juvix
```
module Test;
import Stdlib.Prelude open;
```
When the module is compiled, we have a step in the compiler pipeline
which filters out unreachable symbols. For this module all symbols are
filtered because the module contains no definitions.
So if the module is loaded in the REPL, no symbols will be available to
process through the evaluator. The REPL is a place to explore the
symbols in the module so (like with Haskell's GHCi) it would be useful
if all symbols were available in the REPL session. That's what this PR
implements.
* Closes https://github.com/anoma/juvix/issues/2159
---------
Co-authored-by: Jan Mas Rovira <janmasrovira@gmail.com>
- Closes#2128
- Closes#2161
This pr fully implements the monadic pretty printer based on
`ExactPrint`, which respects comments. Until now, comments inside
expressions were printed after the current statement. Now they are
printed in the correct place, except when a comment occurs before
something that we don't store its location. E.g. parentheses,
semicolons, braces, colons, etc. I proposed that we irone out this issue
in a separate pr.
Since the old non-monadic algorithm is no longer necessary, I removed
it.
* Closes#2134
Adds the `argnames` pragma which specifies function argument names.
These will be the names used in Core and subsequently in VampIR for the
`main` function.
```
{-# argnames: [x, y] -#}
main : Nat -> Nat -> Nat;
```
`AnsiText` is a type that represents some text that can be printed with
`Ansi` formatting annotations, or as plain text. It is expected that it
should have a `Semigroup` instance. This pr adds that.
---------
Co-authored-by: Paul Cadman <git@paulcadman.dev>
This pr adds a new command, `:def` to the repl. This command expects a
single identifier that must be in scope and then prints its definition.
For constructors, the whole type definition is printed.
It also applies some refactors to the code for repl command.
1. Before there was a mega `where` block of definitions. I have hoisted
most of the definitions there to the top level. I feel like now it is
easier to navigate and read.
2. Use `ExceptT` instead of local `case` expressions for errors.
3. Use forks of `haskeline` and `repline`. These forks are necessary
because these libraries do not export the constructors `HaskelineT` and
`InputT` respectively, thus, making it impossible to catch errors in
their underlying monad.
This PR adds support for importing modules from within a Juvix project
in the Juvix REPL.
The imported module is checked (parsed, arity-checked, type-checked etc)
as normal and added to the REPL session scope. Any errors during the
checking phase is reported to the user.
### Notes:
* You must load a file before using `import`. This is because the REPL
needs to know which Juvix project is active.
* You may only import modules from within the same Juvix project.
### Examples
After launching `juvix repl`:
#### `open import`
```
Stdlib.Prelude> open import Stdlib.Data.Int.Ord
Stdlib.Prelude> 1 == 1
true
```
#### `import as`
```
Stdlib.Prelude> import Stdlib.Data.Int.Ord as Int
Stdlib.Prelude> 1 Int.== 1
true
```
#### `import`then `open`
```
Stdlib.Prelude> import Stdlib.Data.Int.Ord as Int
Stdlib.Prelude> open Int
Stdlib.Prelude> 1 == 1
true
```
#### Line-terminating semicolons are ignored:
```
Stdlib.Prelude> import Stdlib.Data.Int.Ord as Int;;;;;
Stdlib.Prelude> 1 Int.== 1
true
```
* Closes https://github.com/anoma/juvix/issues/1951
---------
Co-authored-by: Jonathan Cubides <jonathan.cubides@uib.no>
This pr fixes a bug where the repl would crash if it had the implicit
stdlib dependency and the .juvix-build/stdlib directory did not yet
exist. This bug was not exposed in the smoke tests because the
.juvix-build was never cleared.
---------
Co-authored-by: Paul Cadman <git@paulcadman.dev>
# Description
No the filepath in the `juvix forma` command is n=made optional.
However, in that case, the `--stdin` command is required.
### Implementation details
~For now, as a quick solution, I have introduce the "fake" path that is
used for `fomat` command with stdin option.~
I needed to do a couple of big changes:
* `format` command FILE is now optional, howvere, I check that in case
of `Nothing` `--stdin` option should be present, otherwise it will fail
* `entryPointModulePaths` is now `[]` instead of `NonEmpty`
* `ScopeEff` now has `ScopeStdin` constructor as well, which would take
the input from stdin instead of having path passed around
* `RunPipelineNoFileEither` is added to the `App` with the bunch of
`*Stdin` functions that doesn't require filepath argument to be passed
Fixes#2008
## Type of change
- [x] New feature (non-breaking change which adds functionality)
# Checklist:
- [x] My code follows the style guidelines of this project
- [x] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my
feature works:
- [x] smoke tests
---------
Co-authored-by: Paul Cadman <git@paulcadman.dev>
Co-authored-by: janmasrovira <janmasrovira@gmail.com>
* Closes#1965
* Implements the `unroll` pragma to control the unrolling depth on a
per-function basis.
* Implements parsing of the `inline` pragma.
---------
Co-authored-by: janmasrovira <janmasrovira@gmail.com>
- Closes#1993
This pr makes it possible to use `~`, `..` and environment variables in
the `juvix.yaml` and all flags / input of the cli.
In the CLI, the shell will be responsible for replacing environment
variables with their value, so the usual syntax can be used. For the
`dependencies` field, I have implemented a parser that has some
restrictions:
1. Environment variables are given with the makefile-like syntax
`$(VAR)`
2. The three characters `$` `(` `)` are reserved for the environment
variables syntax.
They cannot be part of the path.
3. `~` is reserved for `$(HOME)`. I.e. the prepath `~~` will expand to
`$HOME$HOME`.
4. Nested environment variables are not allowed.
Thanks @paulcadman for the feedback. I think we are ready to merge this
nightmarish pr 👻
---------
Co-authored-by: Paul Cadman <git@paulcadman.dev>
This PR implements pretty printing of evaluation results consistently
with Juvix syntax. The printed values do not necessarily originate
directly from the source code. All functions/lambdas are printed as
`<fun>`.
The same mechanism is used to implement pretty printing of unmatched
pattern examples.
Juvix REPL now uses the new value printing mechanism to display
evaluation results. Typing `nil` in the REPL will now just display
`nil`. The command `juvix dev repl` still prints raw JuvixCore terms.
* Closes#1957
* Closes#1985
There were two bugs here:
1. The _replNoDisambiguate option was set to True for `juvix repl`.
2. The disambiguateNames pass was being applied after the result node
had already been fetched from the infotable.
* Closes https://github.com/anoma/juvix/issues/1958
The new `juvix dev repl` command is a copy of the `juvix repl` with the
addition of `--no-disambiguate` flag that is present on the `juvix dev
core from-concrete` command.
The `juvix repl` command now does not have the `--transforms`,
`--show-de-bruijn` flags as these are only relevant for compiler
developers. The eval transforms are always applied.
By default `juvix dev repl` uses the eval transforms. You can override
this by specifying the `-t` flag.
Also we now run `disambiguateNames` transform on the info table in the
`dev repl` (unless the `--no-disambiguate-names` flag is set). This is
so the output of the `juvix dev repl` will match that of `juvix dev core
from-concrete` and also so the output can be parsed by back the core
parser.
* Closes https://github.com/anoma/juvix/issues/1914
Previously we were:
* discarding the types table
* discarding the name ids state
after processing an expression in the REPL.
For example evaluating:
```
let even : _; odd : _; odd zero := false; odd (suc n) := not (even n); even zero := true; even (suc n) := not (odd n) in even 10
```
would loop in the REPL.
We noticed that the `n` in `suc n` was being given type `Type` instead
of `Nat`. This was because the name id given to n was incorrect, the
REPL started using name ids from 0 again.
We fixed this issue by storing information, including the types table
and name ids state in the Artifacts data structure that is returned when
we run the pipeline for the first time. This information is then used
when we call functions to compile / type check REPL expressions.
---------
Co-authored-by: Paul Cadman <git@paulcadman.dev>
This implements a basic version of the algorithm from: Luc Maranget,
[Compiling pattern matching to good decision
trees](http://moscova.inria.fr/~maranget/papers/ml05e-maranget.pdf). No
heuristics are used - the first column is always chosen.
* Closes#1798
* Closes#1225
* Closes#1926
* Adds a global `--no-coverage` option which turns off coverage checking
in favour of generating runtime failures
* Changes the representation of Match patterns in JuvixCore to achieve a
more streamlined implementation
* Adds options to the Core pipeline
* Depends on #1832
* Closes#1844
* Adds errors to the Core pipeline
* Checks for no recursion in the GEB pipeline
* Checks for no polymorphism in the GEB pipeline
* Checks for no dynamic type in the GEB pipeline
* Checks for no IO in the GEB pipeline
* Checks for no unsupported builtins in the GEB pipeline
* Depends on PR #1824
* Closes#1556
* Closes#1825
* Closes#1843
* Closes#1729
* Closes#1596
* Closes#1343
* Closes#1382
* Closes#1867
* Closes#1876
* Changes the `juvix compile` command to use the new pipeline.
* Removes the `juvix dev minic` command and the `BackendC` tests.
* Adds the `juvix eval` command.
* Fixes bugs in the Nat-to-integer conversion.
* Fixes bugs in the Internal-to-Core and Core-to-Core.Stripped
translations.
* Fixes bugs in the RemoveTypeArgs transformation.
* Fixes bugs in lambda-lifting (incorrect de Bruijn indices in the types
of added binders).
* Fixes several other bugs in the compilation pipeline.
* Adds a separate EtaExpandApps transformation to avoid quadratic
runtime in the Internal-to-Core translation due to repeated calls to
etaExpandApps.
* Changes Internal-to-Core to avoid generating matches on values which
don't have an inductive type.
---------
Co-authored-by: Paul Cadman <git@paulcadman.dev>
Co-authored-by: janmasrovira <janmasrovira@gmail.com>
Core transformations apply to the whole InfoTable, the REPL needs to
apply Core transformations to the single node that it compiles from the
user input string.
The solution in this commit is to:
1. Compile the input string as before to obtain a Core Node.
2. Add this Node to a copy of the Core InfoTable for the loaded file.
3. Apply the (CLI specified) Core transformations to this InfoTable.
4. Extract the (now transformed) Node from the InfoTable.
We can think of a way to improve this, maybe when we tackle allowing the
user to make new definitions in the REPL.
As soon as compilation of pattern matching is complete we should enable
some (all?) Core transformations by default.
Example:
At the moment we get the following result in the REPL:
```
juvix repl
...
Stdlib.Prelude> 1 + 1
suc (suc zero)
```
After this commit we can turn on `nat-to-int` transformation:
```
juvix repl -t nat-to-int
Stdlib.Prelude> 1 + 1
2
```
* Part of https://github.com/anoma/juvix/issues/1531
Filepaths within a Loc must now be absolute or an error is thrown when
mkLoc is called. This Loc is used when displaying errors.
This commit uses imaginary absolute file paths in the Core repl and Asm
commands in the cases (parsing a single expression for example).
Before this fix, the `core {repl, read, eval}` and `asm` commands would
crash if it encountered an error when invoked with a relative path, or
in the case of a repl when parsing a single expression.
This PR adds some maintenance at different levels to the CI config, the
Make file, and formatting.
- Most of the actions used by the CI related to haskell, ormolu, hlint
and pre-commit have been updated because Github requires NodeJS 16. This
change removes all the old warnings related to nodeJs.
In the case of ormolu, the new version makes us format some files that
were not formatted before, similarly with hlint.
- The CI has been updated to use the latest version of the Smoke testing
framework, which introduced installation of the dependencies for Linux
(libicu66) and macOS (icu4c) in the CI. In the case of macOS, the CI
uses a binary for smoke. For Linux, we use stack to build smoke from the
source. The source here is in a fork of [the official Smoke
repo](https://github.com/SamirTalwar/smoke). Such includes some
features/changes that are not yet in the official repo.
- The Makefile runs the ormolu and hlint targets using as a path for the
binaries the environment variables ORMOLU and HLINT. Thus, export those
variables in your environment before running `make check,` `make format`
or `make hlint`. Otherwise, the Makefile will use the binaries provided
by `stack`.
Co-authored-by: Paul Cadman <git@paulcadman.dev>
* Add shelltests for `juvix repl`
* repl: Compute entrypoint root when loading a file
Previously the REPL would use app root (which could be the current
directory or the project root of the initially loaded file). Thus files
in a different project could not be loaded.
The entrypoint root must be computed each time a new file is `:load`ed.
Adds shell-tests for REPL commands
* Move preludePath to Juvix.Extra.Paths
* Adds a major mode for juvix-repl
* juvix-mode: C-c C-l loads file into REPL, if the REPL is running
* Detect ANSI support in Emacs REPL / shell by using hSupportsANSIColor API
* Disable comint echo handling. Juvix REPL does not echo
* repl: whitespace strip input to compile/eval/infer commands
* REPL support reloading
* Reformat :help and add :version command
* repl: Display the currently loaded module in the banner
* repl: Load the Prelude at launch unless --no-prelude or --no-stdlib are set
