--- /dev/null
+<!--
+Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
+
+SPDX-License-Identifier: curl
+-->
+
+# Infrastructure in the curl project
+
+Overview of infrastructure we maintain, host and run in the project for the
+project.
+
+## git repository
+
+Since 2010, the main curl git repository has been hosted by GitHub, available
+at https://github.com/curl/curl.
+
+We also use the issue tracker, pull requests and discussions on GitHub.
+
+curl has an "enterprise account" on GitHub and is an "organization" on the
+site.
+
+We accept sponsorship via GitHub Sponsors.
+
+## CI services
+
+For every pull request and git push to the master repository, a number of
+build and testing jobs are run on a set of different CI services. The exact
+services vary over time. GitHub Actions and AppVeyor are the primary ones
+these days.
+
+## Test Clutch
+
+A [Test Clutch](https://github.com/dfandrich/testclutch) instance generates
+regular reports on curl CI test results at https://testclutch.curl.se/ as well
+as writing comments on curl pull requests whose tests have failed. The jobs
+are hosted on a Virtuozzo Application Platform PaaS instance and is managed by
+Dan Fandrich. The configuration code is is available and managed at
+https://github.com/dfandrich/testclutch-curl-web
+
+## Autobuilds
+
+The curl autobuild system is a set of scripts that build and test curl and
+send all output logs back to the autobuild server. The results are
+continuously collected and visualized on the curl website at
+<https://curl.se/dev/builds.html>.
+
+The autobuild system and server is maintained by Daniel Stenberg.
+
+## OSS-Fuzz
+
+Google runs the [OSS-Fuzz](https://google.github.io/oss-fuzz/) project which
+also runs fuzzing on curl code, non-stop, in their infrastructure and they
+send us emails in the rare instances they actually find something.
+
+OSS-Fuzz notifies those that are members in the "curl team". Any curl
+maintainer who wants to is welcome to participate. It requires a Google
+account.
+
+## Coverity
+
+We regularly run our code through the [Coverity static code
+analyzer](https://scan.coverity.com/) thanks to them offering this service to
+us for free.
+
+## CodeSonar
+
+[CodeSonar](https://codesecure.com/our-products/codesonar/) analyzes the curl
+source code daily and emails Daniel Stenberg whenever it finds suspected
+problems in the source code. I hope and expect that we can invite other
+maintainers to access these reports soon.
+
+## Domain names
+
+The project runs services and website using a few different curl related
+domain names, including `curl.se` and `curl.dev`. Daniel Stenberg owns these
+domain names.
+
+Until a few years ago, the curl website was present at `curl.haxx.se`. The
+`haxx.se` domain is owned by Haxx AB, administrated by Daniel Stenberg. The
+curl.haxx.se name is meant to keep working and be redirecting to curl.se for
+the foreseeable future.
+
+## Websites
+
+The main curl website at `curl.se` is maintained by curl maintainers and the
+content is available and managed at https://github.com/curl/curl-www. The site
+updates from git and runs make every 20 minutes. Any change pushed to git can
+thus take up to 20 minutes until it takes effect on the origin server.
+
+The content on `curl.dev` is available and managed at
+https://github.com/curl/curl.dev/
+
+The content on `everything-curl.dev` is available and managed at
+https://github.com/curl/everything-curl/
+
+The machine hosting the website contents for these three sites is owned by
+Haxx AB and is primarily managed by Daniel Stenberg (co-owner of the Haxx
+company). The machine is physically located in Sweden.
+
+curl release tarballs are hosted on https://curl.se/download.html. They are
+uploaded there at release-time by the release manager.
+
+curl-for-win downloads are hosted on https://curl.se/windows and are uploaded
+to the server by Viktor Szakats.
+
+curl-for-QNX downloads are hosted on <https://curl.se/qnx> and are uploaded to
+the server by Daniel Stenberg.
+
+Daily release tarball-like snapshots are generated automatically and are
+provided for download at <https://curl.se/snapshots/>.
+
+CA certificate bundles are extracted from the Firefox source code, hosted by
+Mozilla and converted to PEM file format and is offered for download. The
+conversion checks for updates daily. The bundle is provided for download at
+<https://curl.se/docs/caextract.html>.
+
+There is an automated "download check bot" that runs twice daily to scan for
+available curl downloads to populate the curl download page appropriately with
+the correct updated information. The bot uses URLs and patterns for all
+download packages and is maintained in a database, maintained by Daniel
+Stenberg and Dan Fandrich.
+
+The TLS certificate for the origin curl web server is automatically updated
+from Let's Encrypt.
+
+## CDN
+
+Fastly runs the Content Delivery Network (CDN) that fronts all the curl
+websites. The CDN caches content that it gets from the origin server.
+Recently, roughly 99.99% of web requests are satisfied by the CDN without
+having to reach the origin.
+
+The CDN caches different content at different lengths depending on the
+content-type. The caching thus adds to the time for a change to have an effect
+on the site from the moment it gets pushed to the git repository.
+
+Using this setup, we provide four IPv4 addresses and eight IPv6 addresses for
+anycast access to the site. Should be snappy from virtually everywhere across
+the globe.
+
+The CDN servers support HTTP/1, HTTP/2 and HTTP/3. They set HSTS for a year.
+The `HTTP://` version of the site redirects to `HTTPS://`.
+
+Fastly manages the TLS certificates from Let's Encrypt for the servers they
+run on the behalf of curl.
+
+## Containers
+
+The curl project offer container builds of curl. The source repository for
+them is located at <https://github.com/curl/curl-container>.
+
+Container images are hosted at <https://quay.io/repository/curl/curl> and
+<https://hub.docker.com/r/curlimages/curl>
+
+## DNS
+
+The primary domain name, `curl.se` is managed by Kirei and is offered over
+fault-tolerant anycast servers. High availability and fast access for
+everyone.
+
+The actual physical DNS files and origin bind instance is managed by Daniel
+Stenberg.
+
+## Mailing lists
+
+The curl related mailing lists are hosted by Haxx AB on `lists.haxx.se` and
+are maintained by Daniel Stenberg. This includes the mailman2 and Postfix
+instances used for this.
+
+## Email
+
+We use a few rare additional curl related email aliases in the curl domains.
+They go through the mail server `mail.haxx.se` maintained by Daniel Stenberg
+
+## Bug-bounty
+
+We run a [bug-bounty](https://curl.se/docs/bugbounty.html) on HackerOne. The
+setup runs entirely at https://hackerone.com/curl.
+
+The money part for the bug bounty is sponsored by the [Internet Bug
+Bounty](https://hackerone.com/ibb).
+
+## Open Collective
+
+We use [Open Collective](https://opencollective.com/curl) as our "fiscal
+host". All money sent to and received by the curl project is managed by Open
+Collective.
+
+## Merchandise
+
+We have stickers, coffee mugs and coasters. They are managed by Daniel who
+sits on the inventory. The best way to get your hands on curl merchandise is
+to attend events where Daniel is physically.
projects / thirdparty / curl.git / commitdiff
