rust/src/libstd/workcache.rs

436 lines
13 KiB
Rust
Raw Normal View History

// Copyright 2012 The Rust Project Developers. See the COPYRIGHT
// file at the top-level directory of this distribution and at
// http://rust-lang.org/COPYRIGHT.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
#[allow(deprecated_mode)];
2012-11-23 01:24:30 -06:00
use json;
use sha1;
use serialize::{Encoder, Encodable, Decoder, Decodable};
2013-01-07 15:08:43 -06:00
use sort;
use core::cell::Cell;
2013-01-07 15:08:43 -06:00
use core::cmp;
use core::either::{Either, Left, Right};
use core::io;
2013-02-02 05:10:12 -06:00
use core::comm::{oneshot, PortOne, send_one};
use core::pipes::recv;
use core::prelude::*;
use core::result;
use core::run;
use core::hashmap::linear::LinearMap;
use core::task;
use core::to_bytes;
2013-01-07 15:08:43 -06:00
use core::mutable::Mut;
/**
*
* This is a loose clone of the fbuild build system, made a touch more
* generic (not wired to special cases on files) and much less metaprogram-y
* due to rust's comparative weakness there, relative to python.
*
* It's based around _imperative bulids_ that happen to have some function
* calls cached. That is, it's _just_ a mechanism for describing cached
* functions. This makes it much simpler and smaller than a "build system"
* that produces an IR and evaluates it. The evaluation order is normal
* function calls. Some of them just return really quickly.
*
* A cached function consumes and produces a set of _works_. A work has a
* name, a kind (that determines how the value is to be checked for
* freshness) and a value. Works must also be (de)serializable. Some
* examples of works:
*
* kind name value
* ------------------------
* cfg os linux
* file foo.c <sha1>
* url foo.com <etag>
*
* Works are conceptually single units, but we store them most of the time
* in maps of the form (type,name) => value. These are WorkMaps.
*
* A cached function divides the works it's interested up into inputs and
2013-01-07 15:08:43 -06:00
* outputs, and subdivides those into declared (input) works and
* discovered (input and output) works.
*
2013-01-07 15:08:43 -06:00
* A _declared_ input or is one that is given to the workcache before
* any work actually happens, in the "prep" phase. Even when a function's
* work-doing part (the "exec" phase) never gets called, it has declared
2013-01-07 15:08:43 -06:00
* inputs, which can be checked for freshness (and potentially
* used to determine that the function can be skipped).
*
* The workcache checks _all_ works for freshness, but uses the set of
* discovered outputs from the _previous_ exec (which it will re-discover
* and re-record each time the exec phase runs).
*
* Therefore the discovered works cached in the db might be a
* mis-approximation of the current discoverable works, but this is ok for
* the following reason: we assume that if an artifact A changed from
* depending on B,C,D to depending on B,C,D,E, then A itself changed (as
* part of the change-in-dependencies), so we will be ok.
*
* Each function has a single discriminated output work called its _result_.
* This is only different from other works in that it is returned, by value,
* from a call to the cacheable function; the other output works are used in
* passing to invalidate dependencies elsewhere in the cache, but do not
* otherwise escape from a function invocation. Most functions only have one
* output work anyways.
*
* A database (the central store of a workcache) stores a mappings:
*
2013-01-07 15:08:43 -06:00
* (fn_name,{declared_input}) => ({discovered_input},
* {discovered_output},result)
*
2013-01-07 15:08:43 -06:00
* (Note: fbuild, which workcache is based on, has the concept of a declared
* output as separate from a discovered output. This distinction exists only
* as an artifact of how fbuild works: via annotations on function types
* and metaprogramming, with explicit dependency declaration as a fallback.
* Workcache is more explicit about dependencies, and as such treats all
* outputs the same, as discovered-during-the-last-run.)
*
*/
2012-12-11 17:16:36 -06:00
#[deriving_eq]
2013-01-07 15:08:43 -06:00
#[auto_encode]
#[auto_decode]
struct WorkKey {
kind: ~str,
name: ~str
}
impl to_bytes::IterBytes for WorkKey {
#[inline(always)]
pure fn iter_bytes(&self, lsb0: bool, f: to_bytes::Cb) {
let mut flag = true;
self.kind.iter_bytes(lsb0, |bytes| {flag = f(bytes); flag});
if !flag { return; }
self.name.iter_bytes(lsb0, f);
}
}
impl cmp::Ord for WorkKey {
2013-01-07 15:08:43 -06:00
pure fn lt(&self, other: &WorkKey) -> bool {
self.kind < other.kind ||
(self.kind == other.kind &&
self.name < other.name)
}
pure fn le(&self, other: &WorkKey) -> bool {
self.lt(other) || self.eq(other)
}
pure fn ge(&self, other: &WorkKey) -> bool {
self.gt(other) || self.eq(other)
}
pure fn gt(&self, other: &WorkKey) -> bool {
! self.le(other)
}
}
pub impl WorkKey {
static fn new(kind: &str, name: &str) -> WorkKey {
WorkKey { kind: kind.to_owned(), name: name.to_owned() }
}
}
type WorkMap = LinearMap<WorkKey, ~str>;
impl<S:Encoder> Encodable<S> for WorkMap {
2013-01-07 15:08:43 -06:00
fn encode(&self, s: &S) {
2013-02-14 08:12:12 -06:00
let mut d = ~[];
2013-02-07 20:03:13 -06:00
for self.each |&(k, v)| {
2013-01-07 15:08:43 -06:00
d.push((copy *k, copy *v))
}
2013-02-14 08:12:12 -06:00
sort::tim_sort(d);
d.encode(s)
2013-01-07 15:08:43 -06:00
}
}
impl<D:Decoder> Decodable<D> for WorkMap {
2013-01-07 15:08:43 -06:00
static fn decode(&self, d: &D) -> WorkMap {
let v : ~[(WorkKey,~str)] = Decodable::decode(d);
let mut w = LinearMap::new();
2013-02-07 20:03:13 -06:00
for v.each |&(k, v)| {
2013-01-07 15:08:43 -06:00
w.insert(copy k, copy v);
}
w
}
}
struct Database {
2013-01-07 15:08:43 -06:00
db_filename: Path,
db_cache: LinearMap<~str, ~str>,
mut db_dirty: bool
}
pub impl Database {
fn prepare(&mut self,
fn_name: &str,
declared_inputs: &WorkMap) -> Option<(WorkMap, WorkMap, ~str)>
{
2013-01-07 15:08:43 -06:00
let k = json_encode(&(fn_name, declared_inputs));
match self.db_cache.find(&k) {
2013-01-07 15:08:43 -06:00
None => None,
Some(v) => Some(json_decode(*v))
2013-01-07 15:08:43 -06:00
}
}
2013-01-07 15:08:43 -06:00
fn cache(&mut self,
fn_name: &str,
declared_inputs: &WorkMap,
discovered_inputs: &WorkMap,
discovered_outputs: &WorkMap,
result: &str) {
let k = json_encode(&(fn_name, declared_inputs));
let v = json_encode(&(discovered_inputs,
discovered_outputs,
result));
self.db_cache.insert(k,v);
self.db_dirty = true
}
}
struct Logger {
// FIXME #4432: Fill in
a: ()
}
pub impl Logger {
2013-03-07 20:11:09 -06:00
fn info(&self, i: &str) {
2013-01-07 15:08:43 -06:00
io::println(~"workcache: " + i.to_owned());
}
}
struct Context {
2013-01-07 15:08:43 -06:00
db: @Mut<Database>,
logger: @Mut<Logger>,
cfg: @json::Object,
2013-01-07 15:08:43 -06:00
freshness: LinearMap<~str,@fn(&str,&str)->bool>
}
struct Prep {
ctxt: @Context,
fn_name: ~str,
declared_inputs: WorkMap,
}
struct Exec {
discovered_inputs: WorkMap,
discovered_outputs: WorkMap
}
struct Work<T> {
prep: @Mut<Prep>,
res: Option<Either<T,PortOne<(Exec,T)>>>
}
2013-01-07 15:08:43 -06:00
fn json_encode<T:Encodable<json::Encoder>>(t: &T) -> ~str {
do io::with_str_writer |wr| {
t.encode(&json::Encoder(wr));
2013-01-07 15:08:43 -06:00
}
}
fn json_decode<T:Decodable<json::Decoder/&static>>( // FIXME(#5121)
s: &str) -> T
{
2013-01-07 15:08:43 -06:00
do io::with_str_reader(s) |rdr| {
let j = result::unwrap(json::from_reader(rdr));
2013-02-15 01:30:30 -06:00
Decodable::decode(&json::Decoder(j))
2013-01-07 15:08:43 -06:00
}
}
fn digest<T:Encodable<json::Encoder>>(t: &T) -> ~str {
let mut sha = sha1::sha1();
2013-01-07 15:08:43 -06:00
sha.input_str(json_encode(t));
sha.result_str()
}
fn digest_file(path: &Path) -> ~str {
let mut sha = sha1::sha1();
let s = io::read_whole_file_str(path);
sha.input_str(*s.get_ref());
sha.result_str()
}
pub impl Context {
2013-01-07 15:08:43 -06:00
static fn new(db: @Mut<Database>,
lg: @Mut<Logger>,
cfg: @json::Object) -> Context {
2013-01-24 10:47:00 -06:00
Context{db: db, logger: lg, cfg: cfg, freshness: LinearMap::new()}
}
fn prep<T:Owned +
Encodable<json::Encoder> +
Decodable<json::Decoder/&static>>( // FIXME(#5121)
@self,
fn_name:&str,
2013-01-07 15:08:43 -06:00
blk: fn(@Mut<Prep>)->Work<T>) -> Work<T> {
let p = @Mut(Prep {ctxt: self,
fn_name: fn_name.to_owned(),
2013-01-24 10:47:00 -06:00
declared_inputs: LinearMap::new()});
blk(p)
}
}
2013-01-07 15:08:43 -06:00
trait TPrep {
fn declare_input(&self, kind:&str, name:&str, val:&str);
fn is_fresh(&self, cat:&str, kind:&str, name:&str, val:&str) -> bool;
fn all_fresh(&self, cat:&str, map:&WorkMap) -> bool;
fn exec<T:Owned +
Encodable<json::Encoder> +
Decodable<json::Decoder/&static>>( // FIXME(#5121)
&self, blk: ~fn(&Exec) -> T) -> Work<T>;
2013-01-07 15:08:43 -06:00
}
impl TPrep for @Mut<Prep> {
2013-01-07 15:08:43 -06:00
fn declare_input(&self, kind:&str, name:&str, val:&str) {
do self.borrow_mut |p| {
p.declared_inputs.insert(WorkKey::new(kind, name),
val.to_owned());
2013-01-07 15:08:43 -06:00
}
}
2013-01-07 15:08:43 -06:00
fn is_fresh(&self, cat: &str, kind: &str,
name: &str, val: &str) -> bool {
do self.borrow_imm |p| {
let k = kind.to_owned();
let f = (*p.ctxt.freshness.get(&k))(name, val);
2013-01-07 15:08:43 -06:00
do p.ctxt.logger.borrow_imm |lg| {
if f {
lg.info(fmt!("%s %s:%s is fresh",
cat, kind, name));
} else {
lg.info(fmt!("%s %s:%s is not fresh",
cat, kind, name))
}
}
f
}
}
2013-01-07 15:08:43 -06:00
fn all_fresh(&self, cat: &str, map: &WorkMap) -> bool {
2013-02-07 20:03:13 -06:00
for map.each |&(k, v)| {
if ! self.is_fresh(cat, k.kind, k.name, *v) {
return false;
}
}
return true;
}
fn exec<T:Owned +
Encodable<json::Encoder> +
Decodable<json::Decoder/&static>>( // FIXME(#5121)
&self, blk: ~fn(&Exec) -> T) -> Work<T> {
2013-02-15 01:30:30 -06:00
let mut bo = Some(blk);
2013-01-07 15:08:43 -06:00
do self.borrow_imm |p| {
let cached = do p.ctxt.db.borrow_mut |db| {
db.prepare(p.fn_name, &p.declared_inputs)
};
2013-02-15 01:30:30 -06:00
match cached {
2013-01-07 15:08:43 -06:00
Some((ref disc_in, ref disc_out, ref res))
if self.all_fresh("declared input",
2013-01-07 15:08:43 -06:00
&p.declared_inputs) &&
self.all_fresh("discovered input", disc_in) &&
self.all_fresh("discovered output", disc_out) => {
2013-02-15 01:30:30 -06:00
Work::new(*self, Left(json_decode(*res)))
}
2013-01-07 15:08:43 -06:00
_ => {
let (chan, port) = oneshot::init();
let mut blk = None;
blk <-> bo;
let blk = blk.unwrap();
let chan = Cell(chan);
2013-02-15 01:30:30 -06:00
do task::spawn || {
let exe = Exec{discovered_inputs: LinearMap::new(),
discovered_outputs: LinearMap::new()};
let chan = chan.take();
2013-01-07 15:08:43 -06:00
let v = blk(&exe);
2013-02-15 01:30:30 -06:00
send_one(chan, (exe, v));
2013-01-07 15:08:43 -06:00
}
2013-02-15 01:30:30 -06:00
Work::new(*self, Right(port))
2013-01-07 15:08:43 -06:00
}
}
}
}
}
pub impl<T:Owned +
Encodable<json::Encoder> +
Decodable<json::Decoder/&static>> Work<T> { // FIXME(#5121)
2013-01-07 15:08:43 -06:00
static fn new(p: @Mut<Prep>, e: Either<T,PortOne<(Exec,T)>>) -> Work<T> {
2013-02-15 01:30:30 -06:00
Work { prep: p, res: Some(e) }
}
}
// FIXME (#3724): movable self. This should be in impl Work.
fn unwrap<T:Owned +
Encodable<json::Encoder> +
Decodable<json::Decoder/&static>>( // FIXME(#5121)
w: Work<T>) -> T {
2013-02-15 01:30:30 -06:00
let mut ww = w;
let mut s = None;
ww.res <-> s;
2013-02-15 01:30:30 -06:00
match s {
None => fail!(),
2013-02-15 01:30:30 -06:00
Some(Left(v)) => v,
Some(Right(port)) => {
let (exe, v) = match recv(port) {
oneshot::send(data) => data
};
2013-01-07 15:08:43 -06:00
let s = json_encode(&v);
2013-01-07 15:08:43 -06:00
do ww.prep.borrow_imm |p| {
do p.ctxt.db.borrow_mut |db| {
db.cache(p.fn_name,
&p.declared_inputs,
&exe.discovered_inputs,
&exe.discovered_outputs,
s);
}
}
2013-02-15 01:30:30 -06:00
v
}
}
}
2013-01-07 15:08:43 -06:00
//#[test]
fn test() {
use core::io::WriterUtil;
2013-01-07 15:08:43 -06:00
let db = @Mut(Database { db_filename: Path("db.json"),
db_cache: LinearMap::new(),
2013-01-07 15:08:43 -06:00
db_dirty: false });
let lg = @Mut(Logger { a: () });
let cfg = @LinearMap::new();
let cx = @Context::new(db, lg, cfg);
let w:Work<~str> = do cx.prep("test1") |prep| {
let pth = Path("foo.c");
{
let file = io::file_writer(&pth, [io::Create]).get();
2012-12-19 19:28:44 -06:00
file.write_str("int main() { return 0; }");
}
prep.declare_input("file", pth.to_str(), digest_file(&pth));
do prep.exec |_exe| {
let out = Path("foo.o");
run::run_program("gcc", [~"foo.c", ~"-o", out.to_str()]);
2013-02-15 01:30:30 -06:00
out.to_str()
}
};
2013-02-15 01:30:30 -06:00
let s = unwrap(w);
io::println(s);
}