fish-shell/fish-rust/build.rs

198 lines
7.9 KiB
Rust
Raw Normal View History

use rsconf::{LinkType, Target};
use std::error::Error;
fn main() {
cc::Build::new().file("src/compat.c").compile("libcompat.a");
2023-01-15 06:56:24 +08:00
let rust_dir = std::env::var("CARGO_MANIFEST_DIR").expect("Env var CARGO_MANIFEST_DIR missing");
let target_dir =
std::env::var("FISH_RUST_TARGET_DIR").unwrap_or(format!("{}/{}", rust_dir, "target/"));
let fish_src_dir = format!("{}/{}", rust_dir, "../src/");
// Where cxx emits its header.
let cxx_include_dir = format!("{}/{}", target_dir, "cxxbridge/rust/");
// If FISH_BUILD_DIR is given by CMake, then use it; otherwise assume it's at ../build.
let fish_build_dir =
std::env::var("FISH_BUILD_DIR").unwrap_or(format!("{}/{}", rust_dir, "../build/"));
// Where autocxx should put its stuff.
let autocxx_gen_dir = std::env::var("FISH_AUTOCXX_GEN_DIR")
.unwrap_or(format!("{}/{}", fish_build_dir, "fish-autocxx-gen/"));
let mut build = cc::Build::new();
// Add to the default library search path
build.flag_if_supported("-L/usr/local/lib/");
rsconf::add_library_search_path("/usr/local/lib");
let mut detector = Target::new_from(build).unwrap();
// Keep verbose mode on until we've ironed out rust build script stuff
// Note that if autocxx fails to compile any rust code, you'll see the full and unredacted
// stdout/stderr output, which will include things that LOOK LIKE compilation errors as rsconf
// tries to build various test files to try and figure out which libraries and symbols are
// available. IGNORE THESE and scroll to the very bottom of the build script output, past all
// these errors, to see the actual issue.
detector.set_verbose(true);
detect_features(detector);
2023-01-15 06:56:24 +08:00
// Emit cxx junk.
// This allows "Rust to be used from C++"
// This must come before autocxx so that cxx can emit its cxx.h header.
let source_files = vec![
"src/abbrs.rs",
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
"src/ast.rs",
"src/builtins/shared.rs",
"src/common.rs",
"src/env/env_ffi.rs",
"src/env_dispatch.rs",
"src/event.rs",
Port fd_monitor (and its needed components) I needed to rename some types already ported to rust so they don't clash with their still-extant cpp counterparts. Helper ffi functions added to avoid needing to dynamically allocate an FdMonitorItem for every fd (we use dozens per basic prompt). I ported some functions from cpp to rust that are used only in the backend but without removing their existing cpp counterparts so cpp code can continue to use their version of them (`wperror` and `make_detached_pthread`). I ran into issues porting line-by-line logic because rust inverts the behavior of `std::remove_if(..)` by making it (basically) `Vec::retain_if(..)` so I replaced bools with an explict enum to make everything clearer. I'll port the cpp tests for this separately, for now they're using ffi. Porting closures was ugly. It's nothing hard, but it's very ugly as now each capturing lambda has been changed into an explicit struct that contains its parameters (that needs to be dynamically allocated), a standalone callback (member) function to replace the lambda contents, and a separate trampoline function to call it from rust over the shared C abi (not really relevant to x86_64 w/ its single calling convention but probably needed on other platforms). I don't like that `fd_monitor.rs` has its own `c_void`. I couldn't find a way to move that to `ffi.rs` but still get cxx bridge to consider it a shared POD. Every time I moved it to a different module, it would consider it to be an opaque rust type instead. I worry this means we're going to have multiple `c_void1`, `c_void2`, etc. types as we continue to port code to use function pointers. Also, rust treats raw pointers as foreign so you can't do `impl Send for * const Foo` even if `Foo` is from the same module. That necessitated a wrapper type (`void_ptr`) that implements `Send` and `Sync` so we can move stuff between threads. The code in fd_monitor_t has been split into two objects, one that is used by the caller and a separate one associated with the background thread (this is made nice and clean by rust's ownership model). Objects not needed under the lock (i.e. accessed by the background thread exclusively) were moved to the separate `BackgroundFdMonitor` type.
2023-02-18 09:21:44 +08:00
"src/fd_monitor.rs",
2023-01-15 06:56:24 +08:00
"src/fd_readable_set.rs",
Port fd_monitor (and its needed components) I needed to rename some types already ported to rust so they don't clash with their still-extant cpp counterparts. Helper ffi functions added to avoid needing to dynamically allocate an FdMonitorItem for every fd (we use dozens per basic prompt). I ported some functions from cpp to rust that are used only in the backend but without removing their existing cpp counterparts so cpp code can continue to use their version of them (`wperror` and `make_detached_pthread`). I ran into issues porting line-by-line logic because rust inverts the behavior of `std::remove_if(..)` by making it (basically) `Vec::retain_if(..)` so I replaced bools with an explict enum to make everything clearer. I'll port the cpp tests for this separately, for now they're using ffi. Porting closures was ugly. It's nothing hard, but it's very ugly as now each capturing lambda has been changed into an explicit struct that contains its parameters (that needs to be dynamically allocated), a standalone callback (member) function to replace the lambda contents, and a separate trampoline function to call it from rust over the shared C abi (not really relevant to x86_64 w/ its single calling convention but probably needed on other platforms). I don't like that `fd_monitor.rs` has its own `c_void`. I couldn't find a way to move that to `ffi.rs` but still get cxx bridge to consider it a shared POD. Every time I moved it to a different module, it would consider it to be an opaque rust type instead. I worry this means we're going to have multiple `c_void1`, `c_void2`, etc. types as we continue to port code to use function pointers. Also, rust treats raw pointers as foreign so you can't do `impl Send for * const Foo` even if `Foo` is from the same module. That necessitated a wrapper type (`void_ptr`) that implements `Send` and `Sync` so we can move stuff between threads. The code in fd_monitor_t has been split into two objects, one that is used by the caller and a separate one associated with the background thread (this is made nice and clean by rust's ownership model). Objects not needed under the lock (i.e. accessed by the background thread exclusively) were moved to the separate `BackgroundFdMonitor` type.
2023-02-18 09:21:44 +08:00
"src/fds.rs",
2023-01-15 06:56:24 +08:00
"src/ffi_init.rs",
"src/ffi_tests.rs",
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
"src/fish_indent.rs",
"src/future_feature_flags.rs",
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
"src/highlight.rs",
"src/job_group.rs",
2023-05-31 05:22:09 +08:00
"src/kill.rs",
"src/null_terminated_array.rs",
"src/output.rs",
2023-02-05 16:35:06 +08:00
"src/parse_constants.rs",
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
"src/parse_tree.rs",
"src/parse_util.rs",
2023-05-29 09:05:10 +08:00
"src/print_help.rs",
2023-02-04 18:21:42 +08:00
"src/redirection.rs",
"src/signal.rs",
2023-01-15 06:56:24 +08:00
"src/smoke.rs",
"src/termsize.rs",
"src/threads.rs",
2023-02-15 05:54:18 +08:00
"src/timer.rs",
2023-02-05 16:35:06 +08:00
"src/tokenizer.rs",
2023-01-15 06:56:24 +08:00
"src/topic_monitor.rs",
2023-03-28 23:59:51 +08:00
"src/trace.rs",
"src/util.rs",
"src/wait_handle.rs",
2023-01-15 06:56:24 +08:00
];
cxx_build::bridges(&source_files)
2023-01-15 06:56:24 +08:00
.flag_if_supported("-std=c++11")
.include(&fish_src_dir)
.include(&fish_build_dir) // For config.h
.include(&cxx_include_dir) // For cxx.h
.flag("-Wno-comment")
2023-01-15 06:56:24 +08:00
.compile("fish-rust");
// Emit autocxx junk.
// This allows "C++ to be used from Rust."
let include_paths = [&fish_src_dir, &fish_build_dir, &cxx_include_dir];
let mut builder = autocxx_build::Builder::new("src/ffi.rs", include_paths);
// Use autocxx's custom output directory unless we're being called by `rust-analyzer` and co.,
// in which case stick to the default target directory so code intelligence continues to work.
if std::env::var("RUSTC_WRAPPER").map_or(true, |wrapper| {
!(wrapper.contains("rust-analyzer") || wrapper.contains("intellij-rust-native-helper"))
}) {
// We need this reassignment because of how the builder pattern works
builder = builder.custom_gendir(autocxx_gen_dir.into());
}
let mut b = builder.build().unwrap();
2023-01-15 06:56:24 +08:00
b.flag_if_supported("-std=c++11")
.flag("-Wno-comment")
2023-01-15 06:56:24 +08:00
.compile("fish-rust-autocxx");
rsconf::rebuild_if_paths_changed(&source_files);
2023-01-15 06:56:24 +08:00
}
/// Dynamically enables certain features at build-time, without their having to be explicitly
/// enabled in the `cargo build --features xxx` invocation.
///
/// This can be used to enable features that we check for and conditionally compile according to in
/// our own codebase, but [can't be used to pull in dependencies](0) even if they're gated (in
/// `Cargo.toml`) behind a feature we just enabled.
///
/// [0]: https://github.com/rust-lang/cargo/issues/5499
fn detect_features(target: Target) {
for (feature, handler) in [
// Ignore the first entry, it just sets up the type inference. Model new entries after the
// second line.
(
"",
&(|_: &Target| Ok(false)) as &dyn Fn(&Target) -> Result<bool, Box<dyn Error>>,
),
("bsd", &detect_bsd),
("gettext", &have_gettext),
] {
match handler(&target) {
Err(e) => rsconf::warn!("{}: {}", feature, e),
Ok(true) => rsconf::enable_feature(feature),
Ok(false) => (),
}
}
}
/// Detect if we're being compiled for a BSD-derived OS, allowing targeting code conditionally with
/// `#[cfg(feature = "bsd")]`.
///
/// Rust offers fine-grained conditional compilation per-os for the popular operating systems, but
/// doesn't necessarily include less-popular forks nor does it group them into families more
/// specific than "windows" vs "unix" so we can conditionally compile code for BSD systems.
fn detect_bsd(_: &Target) -> Result<bool, Box<dyn Error>> {
// Instead of using `uname`, we can inspect the TARGET env variable set by Cargo. This lets us
// support cross-compilation scenarios.
let mut target = std::env::var("TARGET").unwrap();
if !target.chars().all(|c| c.is_ascii_lowercase()) {
target = target.to_ascii_lowercase();
}
let result = target.ends_with("bsd") || target.ends_with("dragonfly");
#[cfg(any(
target_os = "dragonfly",
target_os = "freebsd",
target_os = "netbsd",
target_os = "openbsd",
))]
assert!(result, "Target incorrectly detected as not BSD!");
Ok(result)
}
/// Detect libintl/gettext and its needed symbols to enable internationalization/localization
/// support.
fn have_gettext(target: &Target) -> Result<bool, Box<dyn Error>> {
// The following script correctly detects and links against gettext, but so long as we are using
// C++ and generate a static library linked into the C++ binary via CMake, we need to account
// for the CMake option WITH_GETTEXT being explicitly disabled.
rsconf::rebuild_if_env_changed("CMAKE_WITH_GETTEXT");
if let Some(with_gettext) = std::env::var_os("CMAKE_WITH_GETTEXT") {
if with_gettext.eq_ignore_ascii_case("0") {
return Ok(false);
}
}
// In order for fish to correctly operate, we need some way of notifying libintl to invalidate
// its localizations when the locale environment variables are modified. Without the libintl
// symbol _nl_msg_cat_cntr, we cannot use gettext even if we find it.
let mut libraries = Vec::new();
let mut found = 0;
let symbols = ["gettext", "_nl_msg_cat_cntr"];
for symbol in &symbols {
// Historically, libintl was required in order to use gettext() and co, but that
// functionality was subsumed by some versions of libc.
if target.has_symbol_in::<&str>(symbol, &[]) {
// No need to link anything special for this symbol
found += 1;
continue;
}
for library in ["intl", "gettextlib"] {
if target.has_symbol(symbol, library) {
libraries.push(library);
found += 1;
continue;
}
}
}
match found {
0 => Ok(false),
1 => Err(format!("gettext found but cannot be used without {}", symbols[1]).into()),
_ => {
rsconf::link_libraries(&libraries, LinkType::Default);
Ok(true)
}
}
}