fix: require an explicit opt-in before rebuilding a source from one day

FileNotFoundError means no recent release carries the asset, which is not the same as the
source never having published: a rate limit or a run of releases missing the asset reaches
the same branch and would erase the accumulated history.

- add --allow-bootstrap; without it a missing asset is now a hard failure
- plumb it through the reusable workflow, and set it for Transport Canada only, which has
  genuinely never published
- let a tolerated source that produced nothing upload nothing, rather than failing its leg
  and blocking the join a required source depends on

Generated-by: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Ashley Childress
2026-09-11 11:06:51 -04:00
parent 9ec3c2ca48
commit 0a0a2855f3
3 changed files with 32 additions and 4 deletions
+10 -2
View File
@@ -23,6 +23,11 @@ on:
required: false
type: boolean
default: false
allow_bootstrap:
description: 'Onboarding only: permit a single-day rebuild when no asset has ever been published'
required: false
type: boolean
default: false
# Keyed on the source: without it every matrix leg shares one group and the legs cancel
# each other, which is the opposite of running them in parallel.
@@ -70,8 +75,9 @@ jobs:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
SOURCE: ${{ inputs.source }}
RUN_DATE: ${{ inputs.date }}
BOOTSTRAP: ${{ inputs.allow_bootstrap && '1' || '' }}
run: |
python "src/create_daily_${SOURCE}_release.py" ${RUN_DATE:+--date "$RUN_DATE"}
python "src/create_daily_${SOURCE}_release.py" ${RUN_DATE:+--date "$RUN_DATE"} ${BOOTSTRAP:+--allow-bootstrap}
# Stage into one directory so every leg's artifact has the same root; a second
# search path would move the root to the common ancestor for some legs only.
shopt -s nullglob
@@ -93,4 +99,6 @@ jobs:
name: registry-${{ inputs.source }}
path: data/registry-out
retention-days: 1
if-no-files-found: error
# A tolerated source that failed has nothing to upload; only a required
# source missing its artifact is an error.
if-no-files-found: ${{ inputs.required && 'error' || 'ignore' }}
+11 -1
View File
@@ -4,6 +4,10 @@ import argparse
parser = argparse.ArgumentParser(description="Create daily FAA release")
parser.add_argument("--date", type=str, help="Date to process (YYYY-MM-DD format, default: today)")
parser.add_argument("--allow-bootstrap", action="store_true",
help="Permit rebuilding from a single day when no published asset is found. "
"Onboarding only: a missing asset is otherwise indistinguishable from a "
"transient outage, and rebuilding would erase the accumulated history.")
args = parser.parse_args()
if args.date:
@@ -43,7 +47,13 @@ df_new = convert_faa_master_txt_to_df(zip_path, date_str)
try:
df_base, start_date_str = get_latest_aircraft_faa_csv_df()
except FileNotFoundError as e:
print(f"No existing FAA release found, bootstrapping from today only: {e}")
if not args.allow_bootstrap:
raise SystemExit(
f"No published FAA asset found: {e}\n"
"This is indistinguishable from a transient outage, and rebuilding from one day "
"would erase the accumulated history. Pass --allow-bootstrap when onboarding."
) from None
print(f"Bootstrapping FAA from today only (--allow-bootstrap): {e}")
df_base = None
start_date_str = date_str
+11 -1
View File
@@ -4,6 +4,10 @@ import argparse
parser = argparse.ArgumentParser(description="Create daily Transport Canada release")
parser.add_argument("--date", type=str, help="Date to process (YYYY-MM-DD format, default: today)")
parser.add_argument("--allow-bootstrap", action="store_true",
help="Permit rebuilding from a single day when no published asset is found. "
"Onboarding only: a missing asset is otherwise indistinguishable from a "
"transient outage, and rebuilding would erase the accumulated history.")
args = parser.parse_args()
if args.date:
@@ -57,7 +61,13 @@ df_new = convert_tc_ccarcs_to_df(zip_path, date_str)
try:
df_base, start_date_str = get_latest_aircraft_tc_csv_df()
except FileNotFoundError as e:
print(f"No existing Transport Canada release found, bootstrapping from today only: {e}")
if not args.allow_bootstrap:
raise SystemExit(
f"No published Transport Canada asset found: {e}\n"
"This is indistinguishable from a transient outage, and rebuilding from one day "
"would erase the accumulated history. Pass --allow-bootstrap when onboarding."
) from None
print(f"Bootstrapping Transport Canada from today only (--allow-bootstrap): {e}")
df_base = None
start_date_str = date_str