Skip to content

Spec 010: Story 6 — analyses wait their turn - #538

Merged
adamjohnwright merged 1 commit into
mainfrom
spec/010-analysis-queue
Oct 19, 2026
Merged

adamjohnwright merged 1 commit into
mainfrom
spec/010-analysis-queue

Conversation

@adamjohnwright

Copy link
Copy Markdown
Contributor

Spec only (specs/010-java-services). Decided 19 Oct 2026: the AnalysisService stays Java, and a small gateway in this repository admits a few computations at a time in front of it, passing everything else through unchanged.

Measured on this host (research R12), with the site's example data and nothing sent outside:

  • Alone: 0.25 s (demo), 2 s (2.6 MB tissue set), 4.4 s with interactors.
  • Twelve at once: 10.5–12.6 s each. Nothing waits and nothing is refused; everything just finishes late. A large enough burst would take every analysis in it past the edge's 100 s limit, so all of them would fail together.

Proposed settings: 4 analyses at a time, a waiting list of 40, a longest wait of 60 s, then 503 Retry-After. The PDF report gets its own lane, 1 at a time.

Adds Story 6, FR-021–024, SC-008, R12 and Phase 8 (T059–T064).

🤖 Generated with Claude Code

Decided 19 Oct 2026: the AnalysisService stays Java, and a small gateway
in this repository admits a few computations at a time in front of it.
Measured on this host with the site's example data: twelve analyses at
once took 10.5-12.6 s each against 2 s alone -- nothing waits or is
refused, all finish late, so a large burst fails together past the edge's
100 s. Story 6, FR-021-024, SC-008, research R12 and tasks T059-T064.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@adamjohnwright
adamjohnwright merged commit da3f39d into main Oct 19, 2026
7 checks passed
@adamjohnwright
adamjohnwright deleted the spec/010-analysis-queue branch October 19, 2026 19:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant