Struct RunQueuesData

pub struct RunQueuesData<'run_queues> { /* private fields */ }

A loaded copy of the on-disk data, for on-the-fly indexing/multiple traversal

Implementations

impl<'run_queues> RunQueuesData<'run_queues>

pub fn run_queues(&self) -> &'run_queues RunQueues

For things to be run from e.g. run_job (e.g. regenerate_index_files function)

pub fn jobs_by_commit_id(&self, commit_id: &GitHash) -> &[(usize, usize)]

Values are (index in pipeline_data, index within its queue_data)

pub fn have_job_with_commit_id(&self, commit_id: &GitHash) -> bool

pub fn run_queue_with_nexts<'s>( &'s self, ) -> impl Iterator<Item = RunQueueDataWithNext<'run_queues, 'run_queues, 's>>

The RunQueueDatas paired with their successor (still in the original, configured, order)

pub fn run_next_job<'s, 'conf, 'r, 'rc>( &'s self, job_runner: JobRunner<'_>, run_context: &mut RunContext, ) -> Result<Option<(&'s BenchmarkingJob, JobStatus)>>

Run the first or most prioritized job in the queues. Returns the job and its status after running it if one was found, None if all runnable queues are empty.

This method needs to be run in a loop forever for daemon style processing. The reason this doesn’t do the looping inside is to allow for a reload of the queue config and then queues. current_stop_start, if given, represents an active stop_start command that was run with stop and now needs a start when the next running action does not require the same command to be stoped. Likewise, this method returns the active stop_start command, if any, by the time it returns. The caller should pass that back into this method on the next iteration. Using SliceOrBox to allow carrying it over a config reload. now should be the current time (at least is understood as such), get it via get_now_chrono() right before calling this method.