Should we cache result of finding completions for ACR registry URIs?
- Dominant language
- Bicep
- Stars
- 3.6k
- Forks
- 830
- Avg merge
- 1d 21m
- Merged PRs (30d)
- 79
Description
If the flag to include all accessible ACR registries in completions is turned on (default is off), then whenever user tries to get completions for "br:", we will query for all container registries they have access to in Azure (might limit that to selected subscriptions later). Right now this is not cached so we'll do it again next time. On my machine this takes around 3s or so. Of course could be slower.
Should we add caching code? If so, how often do we requery?
PublicRegistryModuleMetadataProvider uses the following strategy:
After failures, exponentially increase retry delay time up to max of 2min
After success, requery in background if it's been longer than 1 hour.
Contributor guide
Research direction
Start by reading PublicRegistryModuleMetadataProvider's retry and background-refresh strategy, then trace the completion path for "br:" when all accessible ACR registries are enabled. Done means the cache behavior and requery policy are decided and documented for this completion lookup.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- azure
- Domain
- cloud, tooling
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100