mirror of https://github.com/denoland/deno.git synced 2024-11-26 16:09:27 -05:00

Kevin (Kun) "Kassimo" Qian f2447f6a23 Add more profiling and debugging guides

2018-12-08 21:34:53 -05:00

16 KiB

Raw Blame History

Deno Documentation

Disclaimer

A word of caution: Deno is very much under development. We encourage brave early adopters, but expect bugs large and small. The API is subject to change without notice.

Bug reports do help!

Install

Deno works on OSX, Linux, and Windows. Deno is a single binary executable. It has no external dependencies.

deno_install provides convenience scripts to download and install the binary.

Using Python:

curl -L https://deno.land/x/install/install.py | python

Or using PowerShell:

iex (iwr https://deno.land/x/install/install.ps1)

Note: Depending on your security settings, you may have to run Set-ExecutionPolicy RemoteSigned -Scope CurrentUser first to allow downloaded scripts to be executed.

Deno can also be installed manually, by downloading a tarball or zip file at github.com/denoland/deno/releases. These packages contain just a single executable file. You will have to set the executable bit on Mac and Linux.

Try it:

> deno https://deno.land/thumb.ts

API Reference

To get an exact reference of deno's runtime API, run the following in the command line:

> deno --types

Or see the doc website.

If you are embedding deno in a Rust program, see the rust docs.

Tutorial

An implementation of the unix "cat" program

In this program each command-line argument is assumed to be a filename, the file is opened, and printed to stdout.

import * as deno from "deno";

(async () => {
  for (let i = 1; i < deno.args.length; i++) {
    let filename = deno.args[i];
    let file = await deno.open(filename);
    await deno.copy(deno.stdout, file);
    file.close();
  }
})();

The copy() function here actually makes no more than the necessary kernel -> userspace -> kernel copies. That is, the same memory from which data is read from the file, is written to stdout. This illustrates a general design goal for I/O streams in Deno.

Try the program:

> deno https://deno.land/x/examples/cat.ts /etc/passwd

TCP echo server

This is an example of a simple server which accepts connections on port 8080, and returns to the client anything it sends.

import { listen, copy } from "deno";

(async () => {
  const addr = "0.0.0.0:8080";
  const listener = listen("tcp", addr);
  console.log("listening on", addr);
  while (true) {
    const conn = await listener.accept();
    copy(conn, conn);
  }
})();

When this program is started, the user is prompted for permission to listen on the network:

> deno https://deno.land/x/examples/echo_server.ts
deno requests network access to "listen". Grant? [yN] y
listening on 0.0.0.0:8080

For security reasons, deno does not allow programs to access the network without explicit permission. To avoid the console prompt, use a command-line flag:

> deno https://deno.land/x/examples/echo_server.ts --allow-net

To test it, try sending a HTTP request to it by using curl. The request gets written directly back to the client.

> curl http://localhost:8080/
GET / HTTP/1.1
Host: localhost:8080
User-Agent: curl/7.54.0
Accept: */*

It's worth noting that like the cat.ts example, the copy() function here also does not make unnecessary memory copies. It receives a packet from the kernel and sends back, without further complexity.

Linking to third party code

In the above examples, we saw that Deno could execute scripts from URLs. Like browser JavaScript, Deno can import libraries directly from URLs. This example uses a URL to import a test runner library:

import { test, assertEqual } from "https://deno.land/x/testing/testing.ts";

test(function t1() {
  assertEqual("hello", "hello");
});

test(function t2() {
  assertEqual("world", "world");
});

Try running this:

> deno https://deno.land/x/examples/example_test.ts
Compiling /Users/rld/src/deno_examples/example_test.ts
Downloading https://deno.land/x/testing/testing.ts
Downloading https://deno.land/x/testing/util.ts
Compiling https://deno.land/x/testing/testing.ts
Compiling https://deno.land/x/testing/util.ts
running 2 tests
test t1
... ok
test t2
... ok

test result: ok. 2 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

Note that we did not have to provide the --allow-net flag for this program, and yet it accessed the network. The runtime has special access to download imports and cache them to disk.

Deno caches remote imports in a special directory specified by the $DENO_DIR environmental variable. It default to $HOME/.deno if $DENO_DIR is not specified. The next time you run the program, no downloads will be made. If the program hasn't changed, it won't be recompiled either.

But what if https://deno.land/ goes down? Relying on external servers is convenient for development but brittle in production. Production software should always bundle its dependencies. In Deno this is done by checking the $DENO_DIR into your source control system, and specifying that path as the $DENO_DIR environmental variable at runtime.

How do you import to a specific version? Simply specify the version in the URL. For example, this URL fully specifies the code being run: https://unpkg.com/liltest@0.0.5/dist/liltest.js. Combined with the aforementioned technique of setting $DENO_DIR in production to stored code, one can fully specify the exact code being run, and execute the code without network access.

It seems unwieldy to import URLs everywhere. What if one of the URLs links to a subtly different version of a library? Isn't it error prone to maintain URLs everywhere in a large project? The solution is to import and re-export your external libraries in a central package.ts file (which serves the same purpose as Node's package.json file). For example, let's say you were using the above testing library across a large project. Rather than importing "https://deno.land/x/testing/testing.ts" everywhere, you could create a package.ts file the exports the third-party code:

export { test, assertEqual } from "https://deno.land/x/testing/testing.ts";

And throughout project one can import from the package.ts and avoid having many references to the same URL:

import { test, assertEqual } from "./package.ts";

This design circumvents a plethora of complexity spawned by package management software, centralized code repositories, and superfluous file formats.

Useful command line flags

V8 has many many command-line flags, that you can see with --v8-options. Here are a few particularly useful ones:

--async-stack-traces

How to Profile deno

To start profiling,

# Make sure we're only building release.
export DENO_BUILD_MODE=release
# Build deno and V8's d8.
./tools/build.py d8 deno
# Start the program we want to benchmark with --prof
./target/release/deno tests/http_bench.ts --allow-net --prof &
# Exercise it.
third_party/wrk/linux/wrk http://localhost:4500/
kill `pgrep deno`

V8 will write a file in the current directory that looks like this: isolate-0x7fad98242400-v8.log. To examine this file:

D8_PATH=target/release/ ./third_party/v8/tools/linux-tick-processor
isolate-0x7fad98242400-v8.log > prof.log
# on macOS, use ./third_party/v8/tools/mac-tick-processor instead

prof.log will contain information about tick distribution of different calls.

To view the log with Web UI, generate JSON file of the log:

D8_PATH=target/release/ ./third_party/v8/tools/linux-tick-processor
isolate-0x7fad98242400-v8.log --preprocess > prof.json

Open third_party/v8/tools/profview/index.html in your brower, and select prof.json to view the distribution graphically.

To learn more about d8 and profiling, check out the following links:

How to Debug deno

We can use LLDB to debug deno.

lldb -- target/debug/deno tests/worker.js
> run
> bt
> up
> up
> l

To debug Rust code, we can use rust-lldb. It should come with rustc and is a wrapper around LLDB.

rust-lldb -- ./target/debug/deno tests/http_bench.ts --allow-net
# On macOS, you might get warnings like
# `ImportError: cannot import name _remove_dead_weakref`
# In that case, use system python by setting PATH, e.g.
# PATH=/System/Library/Frameworks/Python.framework/Versions/2.7/bin:$PATH
(lldb) command script import "/Users/kevinqian/.rustup/toolchains/1.30.0-x86_64-apple-darwin/lib/rustlib/etc/lldb_rust_formatters.py"
(lldb) type summary add --no-value --python-function lldb_rust_formatters.print_val -x ".*" --category Rust
(lldb) type category enable Rust
(lldb) target create "../deno/target/debug/deno"
Current executable set to '../deno/target/debug/deno' (x86_64).
(lldb) settings set -- target.run-args  "tests/http_bench.ts" "--allow-net"
(lldb) b op_start
(lldb) r

Build Instructions (for advanced users only)

Prerequisites:

To ensure reproducible builds, deno has most of its dependencies in a git submodule. However, you need to install separately:

Rust >= 1.30.0
Node
Python 2. Not 3.
ccache (Optional but helpful for speeding up rebuilds of V8.)
Extra steps for Windows users:
1. Add python.exe to PATH (e.g. set PATH=%PATH%;C:\Python27\python.exe)
2. Get VS Community 2017. Make sure to select the option to install C++ tools and the Windows SDK.
3. Enable Debugging Tools for Windows. Go to Control Panel -> Windows 10 SDK -> Right-Click -> Change -> Change -> Check Debugging Tools for Windows -> Change -> Finish.

Build:

# Fetch deps.
git clone --recurse-submodules https://github.com/denoland/deno.git
cd deno
./tools/setup.py

# Build.
./tools/build.py

# Run.
./target/debug/deno tests/002_hello.ts

# Test.
./tools/test.py

# Format code.
./tools/format.py

Other useful commands:

# Call ninja manually.
./third_party/depot_tools/ninja -C target/debug

# Build a release binary.
DENO_BUILD_MODE=release ./tools/build.py :deno

# List executable targets.
./third_party/depot_tools/gn ls target/debug //:* --as=output --type=executable

# List build configuration.
./third_party/depot_tools/gn args target/debug/ --list

# Edit build configuration.
./third_party/depot_tools/gn args target/debug/

# Describe a target.
./third_party/depot_tools/gn desc target/debug/ :deno
./third_party/depot_tools/gn help

# Update third_party modules
git submodule update

Environment variables: DENO_BUILD_MODE, DENO_BUILD_PATH, DENO_BUILD_ARGS, DENO_DIR.

Internals

Internal: libdeno API.

deno's privileged side will primarily be programmed in Rust. However there will be a small C API that wraps V8 to 1) define the low-level message passing semantics, 2) provide a low-level test target, 3) provide an ANSI C API binding interface for Rust. V8 plus this C API is called "libdeno" and the important bits of the API is specified here: https://github.com/denoland/deno/blob/master/libdeno/deno.h https://github.com/denoland/deno/blob/master/js/libdeno.ts

Internal: Flatbuffers provide shared data between Rust and V8

We use Flatbuffers to define common structs and enums between TypeScript and Rust. These common data structures are defined in https://github.com/denoland/deno/blob/master/src/msg.fbs

Internal: Updating prebuilt binaries

V8 takes a long time to build - on the order of an hour. We use pre-built V8 libraries stored in a Google Storage bucket instead of rebuilding it from scratch each time. Our build system is however setup such that we can build V8 as part of the Deno build if necessary (useful for debugging or changing various configurations in V8, or building the pre-built binaries themselves). To control whether to use a pre-built V8 or not use the use_v8_prebuilt GN argument.

Use tools/gcloud_upload.py to upload new prebuilt files.

Contributing

See CONTRIBUTING.md.

Changelog

2018.11.27 / v0.2.0 / Mildly usable

An intro talk was recorded.

Stability and usability improvements. fetch() is 90% functional now. Basic REPL support was added. Shebang support was added. Command-line argument parsing was improved. A forwarding service https://deno.land/x was set up for Deno code. Example code has been posted to deno.land/x/examples and deno.land/x/net.

The resources table was added to abstract various types of I/O streams and other allocated state. A resource is an integer identifier which maps to some Rust object. It can be used with various ops, particularly read and write.

2018.10.18 / v0.1.8 / Connecting to Tokio / Fleshing out APIs

Most file system ops were implemented. Basic TCP networking is implemented. Basic stdio streams exposed. And many random OS facilities were exposed (e.g. environmental variables)

Tokio was chosen as the backing event loop library. A careful mapping of JS Promises onto Rust Futures was made, preserving error handling and the ability to execute synchronously in the main thread.

Continuous benchmarks were added: https://denoland.github.io/deno/ Performance issues are beginning to be addressed.

"deno --types" was added to reference runtime APIs.

Working towards https://github.com/denoland/deno/milestone/2 We expect v0.2 to be released in last October or early November.

2018.09.09 / v0.1.3 / Scale binding infrastructure

ETA v.0.2 October 2018 https://github.com/denoland/deno/milestone/2

We decided to use Tokio https://tokio.rs/ to provide asynchronous I/O, thread pool execution, and as a base for high level support for various internet protocols like HTTP. Tokio is strongly designed around the idea of Futures - which map quite well onto JavaScript promises. We want to make it as easy as possible to start a Tokio future from JavaScript and get a Promise for handling it. We expect this to result in preliminary file system operations, fetch() for http. Additionally we are working on CI, release, and benchmarking infrastructure to scale development.

2018.08.23 / v0.1.0 / Rust rewrite / V8 snapshot

68d388229e

Complete! https://github.com/denoland/deno/milestone/1

Go is a garbage collected language and we are worried that combining it with V8's GC will lead to difficult contention problems down the road.

The V8Worker2 binding/concept is being ported to a new C++ library called libdeno. libdeno will include the entire JS runtime as a V8 snapshot. It still follows the message passing paradigm. Rust will be bound to this library to implement the privileged part of deno. See deno2/README.md for more details.

V8 Snapshots allow deno to avoid recompiling the TypeScript compiler at startup. This is already working.

When the rewrite is at feature parity with the Go prototype, we will release binaries for people to try.