Skip to content

refactor(cli): rename/rescope ado show subcommands.  #874

@michael-johnston

Description

@michael-johnston

Is your feature request related to a problem? Please describe.

Yes. Several issues make the current CLI harder to use than it needs to be:

  • Overlapping commands: ado show details and ado show summary both expose aggregate information about discovery spaces in overlapping ways (show summary already embeds the same kind of space statistics that show details prints, while show details also prints related resources). For operations, only show details exists—there is no parallel “summary” path—so the mental model is inconsistent across resource types.
  • Confusing naming and scope: Commands like show details, show summary, show entities, show requests, and show results mix “overview,” “tabular measurement/trace data,” and “entity measurement tables” in ways that are easy to confuse. Names like “entities” vs what the command actually outputs (measurement data / timeseries-style tables), and “details” vs “summary,” do not clearly signal what you get or how long it will take.
  • No bulk overview for stats: You cannot easily get the same kind of aggregate stats for many spaces or operations in one go. ado show summary supports multiple space IDs and label/query filters, but ado show details does not, and operation overviews are effectively single-resource only—so comparing or auditing many resources is awkward compared to ado get, which already supports listing and filtering.

Describe the solution you'd like

1. ado show stats

Introduce ado show stats as the single command for aggregate, computed statistics about discovery spaces and operations—analogous in ergonomics to ado get (filter by labels or field queries, pass multiple IDs, optional --use-latest), but returning stats views that may be more expensive to compute than plain resource retrieval.

  • Resource kinds: space and operation.
  • Selection model (like ado get): multiple explicit IDs, and/or --label / -l and/or --query / -q to select resources.
  • Output: Table / CSV / markdown (and existing summary-style formats) for comparing many resources in one run.
  • Separation of concerns: This command should not duplicate ado show related; relationship lists and future tree/deps views stay on ado show related. ado show details / ado show summary can be deprecated or aliased to show stats over time.

2. Merge ado show requests and ado show results into ado show trace (or some other name)

Replace the split between request-level and result-level tables with a single command (e.g. ado show trace) that exposes measurement request/result metadata timeseries views . Support operation and discoveryspace scopes: for a space, aggregate traces across operations that used that space. Use flags such as --requests / --results to select which projection to print, and fold in ado get measurement_request-style single-request inspection where it makes sense (e.g. by request ID).

3. Rename ado show entitiesado show data

Rename the command to data because it surfaces collected (or potential) measurement data in tabular/CSV/JSON form—“data” is clearer than “entities” for most users. Keep entities as a deprecated alias for a transition period.

Describe alternatives you've considered

  • Keep show details and show summary separate: Avoids migration, but preserves overlap and inconsistent space vs operation behavior.
  • Only extend ado get with --stats: Reuses one verb but mixes cheap resource retrieval with heavier aggregations and may confuse defaults and help text.
  • Top-level ado stats / ado trace / ado data instead of ado show …: Flatter CLI, but more root commands; nesting under show groups “views over project data” separately from get (stored resources) and describe (full resource docs).
  • Leave requests/results as two commands: Clearer names for each narrow action, but duplicates setup, help text, and user navigation for what is effectively one “measurement trace” workflow. Also show results is in particular misleading

Additional context

This should be considered along with #736 - c.f. the --stat command proposed there

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions