mirror of https://github.com/pulumi/pulumi.git
ff7337e87a
12496: filestate/projectStore: Optimize ListReferences r=abhinav a=abhinav
ListReferences for project layout is currently inefficient.
It roughly does the equivalent of:
for proj in bucket.List(stacksDir)
for ref in bucket.List(stacksDir + "/" + proj):
# ...
This is fine for local, file-based storage,
but super inefficient for S3 or Azure.
The reason this was written as above is because we treat the bucket
as a directory tree in listBucket.
This changes projectStore.ListReferences to use the underlying
bucket.List method directly--without setting the Delimiter.
This treats the bucket as a flat file list instead of a hierarchy,
and we can extract project and stack names from it in one go.
Co-authored-by: Abhinav Gupta <abhinav@pulumi.com>
|
||
|---|---|---|
| .. | ||
| backend.go | ||
| backend_legacy_test.go | ||
| backend_test.go | ||
| bucket.go | ||
| bucket_test.go | ||
| doc.go | ||
| lock.go | ||
| meta.go | ||
| meta_test.go | ||
| snapshot.go | ||
| stack.go | ||
| state.go | ||
| state_test.go | ||
| store.go | ||
| store_test.go | ||