Merge branch 'release/0.41.0'

Author: zhiayang

.semaphore/semaphore.yml

@@ -5,16 +5,31 @@ agent:
     type: e1-standard-2
     os_image: ubuntu1804
 blocks:
-  - name: "main"
+  - name: "linux-build"
     task:
+      agent:
+        machine:
+          type: e1-standard-2
+          os_image: ubuntu1804
       jobs:
-      - name: main
+      - name: build
         commands:
           - checkout
-          - echo "LLVM 7"
           - sudo echo "deb http://apt.llvm.org/bionic/ llvm-toolchain-bionic-7 main" | sudo tee -a /etc/apt/sources.list
           - sudo echo "deb-src http://apt.llvm.org/bionic/ llvm-toolchain-bionic-7 main" | sudo tee -a /etc/apt/sources.list
           - sudo wget -O - http://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
           - sudo apt -y update
           - sudo apt-get -o Dpkg::Options::="--force-overwrite" --allow-unauthenticated -y install -y llvm-7 llvm-7-dev libllvm7 libmpfr-dev libmpfr6
           - CXX=g++-8 CC=gcc-8 LLVM_CONFIG=llvm-config-7 make -j2 test
+  - name: "macos-build"
+    task:
+      agent:
+        machine:
+          type: a1-standard-4
+          os_image: macos-mojave
+      jobs:
+      - name: build
+        commands:
+          - checkout
+          - HOMEBREW_NO_INSTALL_CLEANUP=1 brew install llvm@7 mpfr libffi
+          - PATH="/usr/local/opt/llvm@7/bin:$PATH" LLVM_CONFIG=llvm-config make -j4 tester

.travis.yml

@@ -9,6 +9,13 @@ matrix:
     - os: osx
       osx_image: xcode9.4
 
+cache:
+    directories:
+    - $HOME/Library/Caches/Homebrew
+
+before_cache:
+    - if [ "$TRAVIS_OS_NAME" == "osx" ]; then brew cleanup; fi
+
 addons:
     apt:
         sources:
@@ -30,8 +37,8 @@ addons:
 
 script:
     - if [ "$TRAVIS_OS_NAME" == "osx" ];
-        then PATH="/usr/local/opt/llvm@7/bin:$PATH" LLVM_CONFIG=llvm-config make tester;
-      else CXX=g++-$GCC_VERSION CC=gcc-$GCC_VERSION LLVM_CONFIG=llvm-config-7 make ci;
+        then PATH="/usr/local/opt/llvm@7/bin:$PATH" LLVM_CONFIG=llvm-config make -j2 tester;
+      else CXX=g++-$GCC_VERSION CC=gcc-$GCC_VERSION LLVM_CONFIG=llvm-config-7 make -j2 ci;
       fi
 
 notifications:

README.md

@@ -1,6 +1,6 @@
 # [Flax](https://flax-lang.github.io)
 
-A low level language with high level syntax and expressibility.
+A low level, general-purpose language with high level syntax and expressibility.
 
 
 [![forthebadge](https://forthebadge.com/images/badges/made-with-crayons.svg)](http://forthebadge.com)
@@ -17,13 +17,10 @@ A low level language with high level syntax and expressibility.
 -----------------------------------------------
 
 
-<p align="center">
+<!-- <p align="center">
   <img src="https://raw.githubusercontent.com/flax-lang/flax/develop/build/d20.gif" />
 </p>
-
-
-
------------------------------------------------
+----------------------------------------------- -->
 
 #### Disclaimer ####
 
@@ -49,12 +46,13 @@ I'm no famous artist but this is my magnum opus, so it'll not be abandoned anyti
 - Operator overloading
 - Generic functions and types
 - Type inference (including for generics)
+- Full compile-time execution (of arbitrary code)
 
 -----------------------------------------------
 
 
 ### Language Syntax
-- See https://flax-lang.github.io (incomplete)
+- See https://flax-lang.github.io (incomplete, outdated, obsolete, etc. etc.)
 
 -----------------------------------------------
 
@@ -96,36 +94,33 @@ do {
 ### Building the Flax compiler
 
 #### Dependencies ####
-- LLVM 7, mostly due to their obsession with changing the IR interface every damn version (6 does not work, 8 does not work)
+- LLVM 7, mostly due to their obsession with changing the IR interface every damn version
 - GMP/MPIR
 - MPFR
+- libffi
 
 
 #### macOS / Linux
 
-- Flax uses a makefile; most likely some form of GNU-compatible `make` will work.
-- LLVM needs to be installed. On macOS, `brew install llvm@7` should work. (note: llvm 8 and up seems to have changed some JIT-related things)
-- For macOS people, simply call `make`.
-- Linux people, call `make linux`.
+- The `makefile` is the preferred way to build on UNIX systems.
+- LLVM needs to be installed. On macOS, `brew install llvm@7` should work, and you might need to do some PPA fiddling for Debian-based distros.
 - A *C++17*-compatible compiler should be used.
 - Find the `flaxc` executable in `build/sysroot/usr/local/bin`
 - Additionally, the (admittedly limited) standard library will be copied from `./libs` to `./build/sysroot/usr/local/lib/flaxlibs/`
 
 
 #### Windows
 
-##### Option 1
-- Install [meson](https://mesonbuild.com/)
-- Edit `meson.build` variables to tell it where to find the libraries -- notably, these are needed: [`libmpir`](http://mpir.org), [`libmpfr`](http://mpfr.org), and most importantly, [`libllvm`](http://llvm.org). Follow the build instructions for each library, preferably generating both Debug and Release *static* libraries.
-- Run `meson build\meson-dbg` (where ever you want, really), followed by `ninja -C build\meson-dbg`.
-- `flaxc.exe` will be in `build\meson-dbg`.
-- Build and profit, hopefully.
+- Install [meson](https://mesonbuild.com/).
+- Run `env MPIR_ROOT_DIR=... LLVM_ROOT_DIR=... meson meson-build-dir` to set the locations for the dependencies (see `meson.build` for the names, there are 4) and configure the build.
+- Optionally, pass `--buildtype=release` to build a release-mode compiler (highly recommended)
+- Run `ninja -C meson-build-dir`.
+- `flaxc.exe` will be found in the build directory.
+
+##### Libraries
+- Download the [prebuilt binaries](https://github.com/flax-lang/flax/releases/tag/win-build-deps) for the 4 dependencies, and place them somewhere.
+- Note: the folder structure of the libraries should be `(llvm|mpir|mpfr|libffi)/(Release|Debug)/(include|lib)/...`
 
-##### Option 2
-- Download the [prebuilt binaries](https://github.com/flax-lang/flax/releases/download/win-build-deps/libraries.zip) for LLVM, MPIR, and MPFR.
-- Set the following environment variables: `DEPS_DBG_INCLUDES_DIR`, `DEPS_DBG_LIBS_DIR`, `DEPS_REL_INCLUDES_DIR`, and `DEPS_REL_LIBS_DIR`. Point them to the location of the libraries you just downloaded. (Look in `appveyor.yml` to see what they should contain -- essentially the include and library paths for `msbuild` to find)
-- Note: the folder structure of the libraries should be `(llvm|mpir|mpfr)/Release/(include|lib)/...`
-- Run `msbuild /p:Configuration=Release`
 
 -----------------------------------------------
 
@@ -142,55 +137,7 @@ do {
 
 ### Contributing
 
-- Found a bug? Want a feature? Just submit a pull request!
-
-
------------------------------------------------
-
-
-### Compiler Architecture
-
-Some stuff about the compiler itself, now. As any noob looking to build a compiler would do, I used the [LLVM Kaleidoscope tutorial](http://llvm.org/docs/tutorial/) as a starting point. Naturally, it being a tutorial did no favours for the code cleanliness and organisation of the Flax compiler.
-
-Flax itself has 4 main passes -- Lexing, Parsing, Typechecking, and finally Codegen. Yes that's right, the shitty typecheck-and-codegen-at-the-same-time architecture of old has been completely replaced!
-
-
-#### Tokenising and Lexing
-
-This isn't terribly complicated. Each file is naturally only looked at once, and a list of tokens and raw lines are stored somewhere. There's always a nagging feeling that token location reporting is flawed, and it probably is.
-
-EDIT: It is.
-
-
-#### Parsing
-
-Broadly speaking this is a recursive descent parser, handwritten. Not terribly efficient, but whatever. This step creates the AST nodes for the next step, although there are some hacks to enable custom operators, which should probably be reimplemented sometime soon.
-
-
-
-#### Typechecking
-
-At this stage, each AST node that the parser produced is traversed, at the file-level. AST nodes are transformed through a typechecking phase into SST nodes (the original meaning of this initialism has been lost). This typechecking consists of solidifying `pts::Type`s into `fir::Type`s; given that the former is simply a stripped-down version of the latter, this is natural.
-
-SST nodes are just AST nodes with refinements; identifiers are resolved to their targets, and function calls also find their intended target here.
-
-Before the rewrite, this used to happen in an intertwined fashion with code generation, which definitely wasn't pretty.
-
-
-
-#### Code Generation
-
-After each file is typechecked, the collector forcibly squishes them together into a single unit, combining the necessary definitions from other files that were imported -- code generation happens at the program level.
-
-During code generation, we output 'Flax Intermediate Representation', or 'FIR'. It's basically a layer on top of LLVM that preserves much of its interface, with some light abstractions built on top. This is where AST nodes actually get transformed into instructions.
-
-Part of the purpose for FIR was to decouple from LLVM, and partly to allow compile-time execution in the future, which would definitely be easier with our own IR (mainly to send and retrieve values across the compiler <> IR boundary).
-
-
-
-#### Translation
-
-After a `fir::Module` is produced by the code generator, we 'translate' this into LLVM code. The entire process can be seen in `source/backend/llvm/translator.cpp`, and clearly we basically cloned the LLVM interface here, which makes it easy to translate into LLVM.
+- Any and all issues and pull requests are welcome!
 
 
 

build/supertiny.flx

@@ -109,10 +109,22 @@ import std::map
 
 
 /*
-	! bugs !
+	! important stuff to fix !
 	{
 		2. splatting tuples into a generic function *will not work*
 		3. pass the overall location to the polymorph solver.
+
+		4. how does named-argument calling interact with varaidic arguments/functions? also, how do
+			optional arguments fit into that picure????
+
+		5. for optional arguments on virtual functions, we need to reach a decision --- do we ban it entirely,
+			or follow what c++ does, which is to use the default values specified on the static type of the pointer?
+			note: we are definitely *not* going to handle it at runtime.
+
+		6. see the error message when we pass a [T1:] to a [T2:...] without splatting; we should probably give a better
+			diagnostic, and eg. recommend splatting it, if the internal types match (ie. T1 == T2).
+
+			at the very least, the current message complains about casting [T1:] to T2, which isn't really correct.
 	}
 
 

build/ultratiny.flx

@@ -3,33 +3,29 @@
 // Licensed under the Apache License Version 2.0.
 
 export ultratiny
+import libc as _
 
-import libc
-import std::math
 
-var value = 401.0
 
-#if value > 30.0
+fn qux(x: str, y: int = 3, args: [str: ...])
 {
-	#run {
-		value = 30
-		libc::printf("true case\n")
+	printf("x = %s, y = %d\n", x, y)
+	for a in args
+	{
+		printf("a: %s\n", a)
 	}
 }
-else
-{
-	#run libc::printf("false case\n")
-}
 
-#run {
-	value = math::sqrt(value)
+fn foo(a: int, b: int, c: int, x: int = 9, y: int = 8, z: int = 7)
+{
+	printf("a = %d, b = %d, c = %d\n", a, b, c)
+	printf("x = %d, y = %d, z = %d\n", x, y, z)
 }
 
 @entry fn main()
 {
-	let k = 30
-	let x = #run math::sqrt(value)
-	libc::printf("sqrt(%.1f) = %.3f\n", value, x)
+	foo(x: 4, 1, 2, y: 5, z: 6, 3)
+	qux(y: 17, "hello, world!", "hi", "my", "name", "is", "bob", "ross")
 }
 
 

external/tinyprocesslib/.gitignore

@@ -98,11 +98,9 @@ Note: this is just a personal log of outstanding issues, shorter rants/ramblings
 	`var foo: SomeStruct` without an initialiser...
 
 
-* ### Foreach loops where you take more than one thing at a time, like this, maybe:
-	`for [ first, middle, last, ... ] in list { ... }`
 
-
-* ### Some kind of construct to make a variable immutable beyond a certain point; thus you could have arbitrarily complex initialisation code,
+* ### Some kind of construct to make a variable immutable beyond a certain point
+	thus you could have arbitrarily complex initialisation code,
 	then have the compiler enforce that your variable is immutable after that point, eg:
 
 	```

external/tinyprocesslib/.travis.yml

@@ -5,10 +5,10 @@
 export libc
 
 // printing
-public ffi fn puts(x: &i8) -> i32
-public ffi fn printf(x: &i8, ...) -> i32
-public ffi fn sprintf(x: &i8, y: &i8, ...) -> i32
-public ffi fn snprintf(x: &i8, l: u64, y: &i8, ...) -> i32
+public ffi fn puts(fmt: &i8) -> i32
+public ffi fn printf(fmt: &i8, ...) -> i32
+public ffi fn sprintf(fmt: &i8, y: &i8, ...) -> i32
+public ffi fn snprintf(fmt: &i8, l: u64, y: &i8, ...) -> i32
 
 public ffi fn fprintf(stream: &void, y: &i8, ...) -> i32
 

issues.md

@@ -0,0 +1,22 @@
+# Flax 1.0 Roadmap
+
+As more of the core features are slowly being finished up, the 1.0 milestone is in sight. This is just a list of things that
+should be finished before the initial implementation can be labelled "fit-for-use". Of course, this doesn't preclude any
+bugfixes that are necessary.
+
+
+## Features
+
+1. Traits
+2. Extensions
+3. Cleanup and modernise operator overloading
+4. Standard library
+5. Proper documentation for the entire language and all features
+6. Proper introduction to language features (eg. for a README)
+
+
+### Cleanup/modernise Operator Overloading
+
+The operator overloading mechanism is dated pre-resolver rewrite, which means that it probably isn't as extensible as we'd
+like it to be. For example, polymorphic operators aren't a possiblity currently, and that should be a thing that is allowed.
+