Shipping applications with the Abyss is a 5 step process:
use camino::Utf8PathBuf;
use gazenot::{AnnouncementKey, Gazenot, ReleaseKey};
#[tokio::main]
async fn main() -> Result<(), miette::Report> {
// Data we want to submit to Abyss
let (source_host, owner) = ("github", "axodotdev");
let apps = vec!["app1", "app2"];
let release_version = "1.0.1".to_owned();
let release_tag = "v1.0.1".to_owned();
let is_prerelease = false;
let announcement_body = "# v1.0.1 (2023-01-01)\n\nWow Cool Changelog".to_owned();
let files = std::collections::HashMap::from([
("app1", vec!["src/lib.rs", "dist-manifest.json"]),
("app2", vec!["src/client.rs", "dist-manifest.json"]),
]);
// Step 1: Create the client
let abyss = Gazenot::into_the_abyss(source_host, owner)?;
// Step 2: Create the Artifact Sets
let packages = apps.into_iter().map(|app| app.to_owned());
let artifact_sets = abyss.create_artifact_sets(packages.clone()).await?;
// Step 3: Upload files
let uploads = artifact_sets.iter().filter_map(|set| {
// Gather up all the files we want to upload to each ArtifactSet
files.get(set.package.as_str()).map(|files| {
let files = files.iter().map(Utf8PathBuf::from).collect();
(set, files)
})
});
abyss.upload_files(uploads).await?;
// Step 4: Create Releases
let releases = artifact_sets.iter().map(|set| {
let release = ReleaseKey {
version: release_version.clone(),
tag: release_tag.clone(),
is_prerelease,
};
(set, release)
});
let _releases = abyss.create_releases(releases).await?;
// Step 5: Announce Releases
//
// You could use the `releases` from the previous step, or upgrade the announcement_sets to
// releases as we do here, for situations where these steps happen on different machines.
let releases = artifact_sets
.iter()
.map(|set| set.to_release(release_tag.clone()))
.collect::<Vec<_>>();
let announcement = AnnouncementKey {
body: announcement_body,
};
abyss.create_announcements(&releases, announcement).await?;
Ok(())
}
Note that in typical usage, steps 2, 3, 4, and 5 would all be done on different machines (as they are different stages of cargo-dist's release pipeline in CI). In that case, you would have a workflow like:
- Plan machine: step 1 + step 2, then serialize and store the ArtifactSets
- Hosting machine: load and deserialize ArtifactSets, then step 1 + step 3
- Publish machine: load and deserialize ArtifactSets, then step 1 + step 4
- Announce machine: load and deserialize ArtifactSets, then step 1 + step 5