Caching v0.3.0~

babyrite uses the concurrent caching library moka to cache channel information retrieved from the Discord API, cutting down on the need to refetch it.

About the caching system's technical constraints

• There is no explicit cache invalidation mechanism.
  • Even after a channel is renamed, deleted, or has its permissions changed, stale data may still be returned until the TTI/TTL expires.
• Cache keys are shared across the entire process.
  • If the caller's guild_id / channel_id match, the cache is reused even for a different guild's Discord link expansion.

Supported features

• Message Preview

GitHub permalinks are not covered — babyrite fetches directly from raw.githubusercontent.com every time.

Cache layout

babyrite builds the following cache layout in the machine's memory.

Cache	Key → Value	Purpose
GUILD_CHANNEL_LIST_CACHE	GuildId → HashMap<ChannelId, GuildChannel>	Per-guild channel list
GUILD_CHANNEL_CACHE	ChannelId → GuildChannel	Individual channels

Both operate with the following shared settings:

• Up to 500 entries
  • The cache is organized on a Least Recently Used (LRU) basis.
  • Entries beyond the size limit are evicted starting with the least recently used data.
• TTI (Time To Idle):
  • Data that hasn't been accessed for 1 hour is automatically removed.
• TTL (Time To Live):
  • Data expires after 12 hours regardless of access frequency.

Cache lookup flow

Message preview

1. Look up GUILD_CHANNEL_CACHE (the individual channel cache)
   • If it's a hit, return it as-is.
2. If not found, look up GUILD_CHANNEL_LIST_CACHE (the channel list cache)
   • If it's a hit, look for the target channel in that map.
   • If it's a miss, fetch from the Discord API and store the result in the cache.
3. If the target channel isn't found in the channel list, search active threads (threads aren't included in the channel list).
4. The channel that's found is ultimately written into the cache as well.