Part of #8811 (production disk-full incident). ## The incident bug If a conan library package download's connection died mid-transfer, `downloadAndExtractPackage()`: - threw an **uncaught exception** (`TypeError: terminated`): the error was emitted on the `Readable.from(res.body)` wrapper, which had no `'error'` listener — `.pipe()` neither attaches nor forwards one; - **never settled its promise**: gunzip/tar saw neither an error nor an end, so neither `resolve` nor `reject` was ever called. This wedged a compilation queue slot forever, which permanently disabled temp dir cleanup (it only runs when the queue is idle) and filled the disk on prod. The download path is rewritten around `stream.pipeline()`, which propagates errors through every stage and guarantees settlement. Per-entry file writes also go through `pipeline()`, so write errors propagate (via `extract.destroy`) instead of being silently dropped, and `next()` fires after the file is fully flushed rather than on stream `'end'`. ## Extraction hardening (from review + an adversarial security pass) - **Zip-slip guard anchored at the per-library extraction root** (`downloadPath/<libId>`; plain `downloadPath` for `extractAllToRoot`): a package can no longer write over a sibling library's files or the compilation's own. The check is `path.relative`-based — immune to prefix collisions (`/tmp/pkg` vs `/tmp/pkg-evil`) and to directories merely *named* with leading dots. - **Only regular-file entries are ever written**: directories, symlinks, hardlinks and other types are drained and skipped, including malformed entries (e.g. a directory claiming non-zero size) whose body could otherwise land on disk as a file. CE never creates links of any kind; prod additionally runs under the `/nosym/tmp` nosymfollow mount. - **Zero-length files extract correctly** (the pre-pipeline code created them as a side effect of an early `createWriteStream` — with a leaked fd; the first pipeline version dropped them entirely). - **Logs preserve stacks** (error objects passed to winston, not interpolated) and the archive-controlled entry name is JSON-stringified. Deliberately out of scope (filed as #8817): decompressed-size caps and `packageUrl` scheme/redirect validation — defense-in-depth against our own conan server, not blockers. See the review-convergence comment below for vectors evaluated and rejected with rationale. ## Tests `test/buildenvsetup-ceconan-tests.ts` (new), against a local HTTP server: - **Severed mid-stream download rejects rather than hanging** — against the pre-fix code this reproduces the production failure exactly (unhandled `TypeError: terminated` + timeout). - Happy-path extraction including a zero-length file. - 404 → rejection. - Zip-slip: full escape and sibling-library escape are skipped; a `..`-named directory inside the library root still extracts. - A malformed sized-directory entry settles (no wedged promise) and writes nothing. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: mattgodbolt-molty <mattgodbolt-molty@users.noreply.github.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Compiler Explorer
Compiler Explorer is an interactive compiler exploration website. Edit code in C, C++, C#, F#, Rust, Go, D, Haskell, Swift, Pascal, ispc, Python, Java, or any of the other 30+ supported languages, and see how that code looks after being compiled in real time.
Bug Report · Compiler Request · Feature Request · Language Request · Library Request · Report Vulnerability
Overview
Multiple compilers are supported for each language, many different tools and visualizations are available, and the UI layout is configurable (thanks to GoldenLayout).
Try out at godbolt.org, or run your own local instance. An overview of what the site lets you achieve, why it's useful, and how to use it is available here, or in this talk.
Compiler Explorer follows a Code of Conduct which aims to foster an open and welcoming environment.
Compiler Explorer was started in 2012 to show how C++ constructs are translated to assembly code. It started as a
tmux session with vi running in one pane and watch gcc -S foo.cc -o - running in the other.
Since then, it has become a public website serving over 3,000,000 compilations per week.
You can financially support this project on Patreon, GitHub, Paypal, or by buying cool gear on the Compiler Explorer store.
Using Compiler Explorer
FAQ
There is now a FAQ section in the repository wiki. If your question is not present, please contact us as described below, so we can help you. If you find that the FAQ is lacking some important point, please feel free to contribute to it and/or ask us to clarify it.
Videos
Several videos showcase some features of Compiler Explorer:
- Compiler Explorer 2023: What's New?: Presentation for CppNorth 2023.
- Presentation for CppCon 2019 about the project
- Older 2 part series of videos which go into a bit more detail into the more obscure features.
- Just Enough Assembly for Compiler Explorer: Practical introduction to Assembly with a focus on the usage of Compiler Explorer, from CppCon 2021.
- Playlist: Compiler Explorer: A collection of videos discussing Compiler Explorer; using it, installing it, what it's for, etc.
A Road map is available which gives a little insight into the future plans for Compiler Explorer.
Developing
Compiler Explorer is written in TypeScript, on Node.js.
Assuming you have a compatible version of node installed, on Linux simply running make ought to get you up and
running with an Explorer running on port 10240 on your local machine:
http://localhost:10240/. If this doesn't work for you, please contact us, as we consider it
important you can quickly and easily get running. Currently, Compiler Explorer requires
node 22 or higher installed, either on the path or at NODE_DIR (an environment variable or
make parameter).
Running with make EXTRA_ARGS='--language LANG' will allow you to load LANG exclusively, where LANG is one for the
language ids/aliases defined in lib/languages.ts. For example, to only run Compiler Explorer with C++ support,
you'd run make EXTRA_ARGS='--language c++'. You can supply multiple --language arguments to restrict to more than
one language. The Makefile will automatically install all the third-party libraries needed to run; using npm to
install server-side and client-side components.
For development, we suggest using make dev to enable some useful features, such as automatic reloading on file changes
and shorter startup times.
You can also use npm run dev to run if make dev doesn't work on your machine.
When making UI changes, we recommend following the UI Testing Checklist to ensure all components work correctly.
Some languages need extra tools to demangle them, e.g. rust, d, or haskell. Such tools are kept separately in the
tools repo.
Configuring compiler explorer is achieved via configuration files in the etc/config directory. Values are key=value.
Options in a {type}.local.properties file (where {type} is c++ or similar) override anything in the
{type}.defaults.properties file. There is a .gitignore file to ignore *.local.* files, so these won't be checked
into git, and you won't find yourself fighting with updated versions when you git pull. For more information see
Adding a Compiler.
Check CONTRIBUTING.md for detailed information about how you can contribute to Compiler Explorer, and the docs folder for specific details regarding various things you might want to do, such as how to add new compilers or languages to the site.
Running a local instance
If you want to point it at your own GCC or similar binaries, either edit the etc/config/LANG.defaults.properties or
else make a new one with the name LANG.local.properties, substituting LANG as needed. *.local.properties files
have the highest priority when loading properties.
For a quick and easy way to add local compilers, use the CE Properties Wizard which automatically detects and configures compilers for 30+ languages. See Adding a Compiler for more details.
If you want to support multiple compilers and languages like godbolt.org, you can use the
bin/ce_install install compilers command in the infra project to install
all or some of the compilers. Compilers installed in this way can be loaded through the configuration in
etc/config/*.amazon.properties. If you need to deploy in a completely offline environment, you may need to remove some
parts of the configuration that are pulled from www.godbolt.ms@443.
When running in a corporate setting the URL shortening service can be replaced by an internal one if the default storage
driver isn't appropriate for your environment. To do this, add a new module in lib/shortener/myservice.js and set the
urlShortenService variable in configuration. This module should export a single function, see the
tinyurl module for an example.
RESTful API
There's a simple restful API that can be used to do compiles to asm and to list compilers.
You can find the API documentation here.
Contact us
We run a Compiler Explorer Discord, which is a place to discuss using or developing
Compiler Explorer. We also have a presence on the cpplang Slack channel
#compiler_explorer and we have
a public mailing list.
There's a development channel on the discord, and also a development mailing list.
Feel free to raise an issue on github or email Matt directly for more help.
Official domains
Following are the official domains for Compiler Explorer:
The domains allow arbitrary subdomains, e.g., https://foo.godbolt.org/, which is convenient since each subdomain has an independent local state. Also, language subdomains such as https://rust.compiler-explorer.com/ will load with that language already selected.
Credits
Compiler Explorer is maintained by the awesome people listed in the AUTHORS file.
We would like to thank the contributors listed in the CONTRIBUTORS file, who have helped shape Compiler Explorer.
We would also like to especially thank these people for their contributions to Compiler Explorer:
- Gabriel Devillers (while working for Kalray)
- Johan Engelen
- Joshua Sheard
- Andrew Pardoe
Many amazing sponsors, both individuals and companies, have helped fund and promote Compiler Explorer.
