Struct WorkflowSpec 

pub struct WorkflowSpec {

    pub entrypoint: String,
    pub templates: Vec<Template>,
    pub arguments: Option<Arguments>,
    pub workflow_template_ref: Option<WorkflowTemplateRef>,
    pub service_account_name: String,
    pub node_selector: BTreeMap<String, String>,
    pub hooks: BTreeMap<String, LifecycleHook>,
    pub ttl_strategy: Option<TtlStrategy>,
    pub pod_gc: Option<PodGc>,
    pub active_deadline_seconds: Option<i64>,
    pub synchronization: Option<Synchronization>,
    pub priority: Option<i32>,
    pub tolerations: Vec<Toleration>,
    pub affinity: Option<Value>,
    pub pod_spec_patch: Option<String>,
    pub image_pull_secrets: Vec<LocalObjectReference>,
    pub parallelism: Option<i64>,
    pub volume_claim_templates: Vec<PersistentVolumeClaim>,
}

Fields

entrypoint: String

templates: Vec<Template>

arguments: Option<Arguments>

workflow_template_ref: Option<WorkflowTemplateRef>

Set on a runnable Workflow that just invokes a WorkflowTemplate.

service_account_name: String

node_selector: BTreeMap<String, String>

Root-scoped pod scheduling for the submitted Workflow (Argo applies it to every pod). Only cargo athena submit --node-selector sets this; emit never does (skip-empty ⇒ existing goldens unaffected).

hooks: BTreeMap<String, LifecycleHook>

Whole-workflow lifecycle hooks. Key exit is the exit handler (runs once when the Workflow finishes). We use this (with a templateRef) rather than the legacy spec.onExit string, which only resolves a local template name — unusable across the one-WT-per-template wormhole.

ttl_strategy: Option<TtlStrategy>

Workflow-scoped TTL GC (#[…(ttl(..))]).

pod_gc: Option<PodGc>

Workflow-scoped pod GC (#[…(pod_gc(strategy=..))]). camelCase of pod_gc is podGc, but Argo’s field is podGC — the argo! macro forwards this explicit rename ahead of its rename_all, so it wins.

active_deadline_seconds: Option<i64>

Root-scoped Argo WorkflowSpec.activeDeadlineSeconds — the genuine whole-workflow runtime cap, from #[…(active_deadline_if_root=..)]. (int64 in Argo; camelCase of the field name already matches.) skip-empty ⇒ existing goldens stay byte-identical.

synchronization: Option<Synchronization>

Root-scoped Synchronization — workflow-level mutexes (#[…(mutexes_if_root = [{ name = … }])]). Argo’s sync manager keys on <ns>/Mutex/<name> globally per controller, so two SEPARATE Workflow runs (not just templateRef’d sub-workflows in one run) contend on the same name — empirically verified on v4.0.5, holder key <ns>/<wf>.

priority: Option<i32>

Workflow priority (int32). The Argo controller schedules higher-priority workflows first when its parallelism limit is hit. Today only cargo athena submit --priority sets this; emit never does (skip-empty ⇒ existing goldens stay byte-identical). A future #[workflow(priority_if_root=N)] would land here too (spec-scoped, hence the _if_root convention).

tolerations: Vec<Toleration>

Root-scoped tolerations (3rd tier of Argo’s tmpl → boundary → wfSpec pod-scheduling lookup). From #[…(tolerations_if_root = [...])]. Applies to every pod in the run that doesn’t have its own template- or boundary-level override (same cascade as node_selector).

affinity: Option<Value>

Root-scoped pod affinity (3rd tier). Opaque YAML/JSON value from #[…(affinity_if_root = "...")]; athena does NOT model the deeply-nested apiv1.Affinity schema by design.

pod_spec_patch: Option<String>

Root-scoped strategic-merge patch applied to every pod in the run after Argo renders the podSpec. From #[workflow(pod_spec_patch_if_root = "...")]. Substituted at pod-creation time (workflow/controller/workflowpod.go:89 processPodSpecPatch → ProcessArgs) so {{workflow.parameters.X}} resolves; per-template Template.pod_spec_patch is concat’d onto this one. The universal escape hatch for any podSpec field athena hasn’t lifted to a first-class attr.

image_pull_secrets: Vec<LocalObjectReference>

Root-only WorkflowSpec.ImagePullSecrets — Secret references the kubelet uses to authenticate against private image registries when pulling each pod’s image. From #[…(image_pull_secrets_if_root = ["regcred", ...])]. K8s/Argo expose this only at workflow scope (no per-template knob); per-container needs go through pod_spec_patch. skip-empty ⇒ existing goldens stay byte-identical.

parallelism: Option<i64>

Root-scoped pod-concurrency cap. From #[workflow(parallelism_if_root = N)]. Argo’s CRD enforces +kubebuilder:validation:Minimum=1 so the macro rejects <= 0 at compile time (would otherwise be rejected by the API server at submit anyway).

volume_claim_templates: Vec<PersistentVolumeClaim>

PVCs Argo creates per workflow run and deletes when it finishes. One entry per #[ephemeral_pvc] type reachable from the submitted root. Each pod mounts them via claimName: <metadata.name>. Root-only by Argo’s design (only the submitted Workflow’s spec is honored).

Trait Implementations

impl Clone for WorkflowSpec

fn clone(&self) -> WorkflowSpec

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for WorkflowSpec

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for WorkflowSpec

fn default() -> WorkflowSpec

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for WorkflowSpec

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for WorkflowSpec

fn eq(&self, other: &WorkflowSpec) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for WorkflowSpec

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for WorkflowSpec