# `mix mishka_cms_core.purge_draft_archival`
[🔗](https://github.com/mishka-group/mishka_cms_core/blob/v0.0.1/lib/mix/tasks/mishka_cms_core.purge_draft_archival.ex#L5)

Deletes runtime drafts that were discarded back when discarding was a soft delete.

`MishkaCmsCore.Runtime.PageDraft` and its two siblings used `AshArchival` until 2026-08-01: a
discard set `archived_at` and a base filter hid the row. Removing the extension took the filter
with it, so those rows became visible again — a draft its author threw away now sits in the picker
looking like work in progress. Run this once, on any database that existed before that date.

`MishkaCmsCore.Runtime.DraftArchivalPurge` holds the logic and explains exactly which rows match
and why a published draft is never one of them.

## Command line options

  * `--dry-run` — report what would be deleted and delete nothing
  * `--quiet` — print nothing; the exit status still reflects success

## Run it BEFORE migrating

The migration that drops `archived_at` destroys the only thing separating a discarded draft from a
live one. After that, no query can tell them apart and this task has nothing to work with — it
reports zero and exits cleanly, which is the right behaviour and also useless.

    mix mishka_cms_core.purge_draft_archival --dry-run
    mix mishka_cms_core.purge_draft_archival
    mix ash.codegen remove_draft_archival
    mix ash.migrate

Running it on a database that never archived, or twice, or after the column is gone, is a clean
no-op in each case.

## In a release

A release has no Mix, and `bin/migrate` drops the column unattended — so the purge has to happen
before it, through `eval`:

    bin/mishka_cms eval 'MishkaCmsCore.Runtime.DraftArchivalPurge.run()'
    bin/migrate

---

*Consult [api-reference.md](api-reference.md) for complete listing*