3 Commits

Include renames

Author	SHA1	Message	Date
sneak	27493568fd	Build the image pipeline in pixad's fx graph (closes #313) ... internal/app now provides the cache, the fetcher, the processor, the image service, the session manager and the encrypted URL generator, and handlers.New receives them built. database.New opens the database, so these parts can read it while fx builds them; the cache's stop hook still runs before the database's. Each setting's default is declared in internal/config, the default cache_max_bytes included, which internal/app works out before it builds the cache. The cache's MaxBytes is the one value for the disk cache limit, and 0 turns the disk cache off. The fetcher and the processor take explicit values. The server tests build the pipeline themselves, as they cannot import internal/app. Model: opus-5-5	2026-10-26 10:31:44 +00:00
clawbot	5b17d1f555	Remove idle host semaphores and delete .meta with its variant (closes #87) ... Each upstream host's semaphore now counts the fetches holding or waiting for one of its slots, and is removed from hostSems when the last of them gives its slot back or stops waiting, so a long-running pixad no longer keeps one semaphore per host it ever fetched from. VariantStorage.Delete removes the variant's .meta file too, a missing one not being an error; DeleteWithMeta, which eviction called for that, is gone. Model: opus-5-5	2026-10-04 07:11:31 +02:00
clawbot	baf457d21e	Bound concurrent image processing and upstream fetches (closes #64) ... Nothing bounded total in-flight work, so a burst of cache misses across hosts could exhaust memory. Two settings now do: max_concurrent_processing (default the CPUs Go uses) and upstream_connections (default 64, beside the per-host limit). A request that finds either full waits up to 10 seconds, then gets 503; a slot is released on every path. No request holds source bytes while it waits: a cached source is read only after the processing slot is taken, and a fetched one only while it holds its upstream connection. libvips runs one worker thread per image with its operation cache off. Both waits count toward downstream_timeout, as the README says. Model: opus-5-5	2026-09-29 10:32:16 +02:00