Release 1.1.0 with adaptive RAM and interactive previews
All checks were successful
Release AstroFrame / packages (push) Successful in 1m13s

Replace fixed process, heap, frame and worker caps with physical RAM admission and CPU-scaled concurrency. Round Windows job budgets to native allocation boundaries. Add persistent cursor-centered zoom, drag and reset, fresh launches, cerulean update indicators and grey native startup backgrounds. Validate with isolated fixtures and a fake updater.
This commit is contained in:
Elijah 2026-10-04 09:33:44 -07:00
parent 6b56c0b122
commit a8e7a87b94
21 changed files with 271 additions and 80 deletions

View file

@ -48,7 +48,7 @@ Small sessions and unreliable references cannot automatically reject. Reference
## Memory, caching and safety
Up to four workers are now selected automatically, leaving two logical processors unused where possible. A 48-frame NGC 7000 measurement benchmark took 9.04/4.99/2.94 seconds with one/two/four workers, with identical star counts and four-worker peak RAM of 250 MiB. This measures the stellar/background stage, not whole-project speedup. No full-resolution image collection or stack is held in RAM. Reference maps are only 960 values per frame. Windows enforces a verified 6 GiB process allocation cap; both executables use a 3 GiB managed-heap cap. A 256 MiB estimated per-frame budget rejects oversized allocations. Analysis checks physical memory before both image passes, reserving up to 8 GiB or one quarter of physical RAM. The page file does not satisfy this reserve. Sustained pressure stops safely and saves measurements.
Version 1.1.0 selects up to 16 workers from CPU capacity, leaving two logical processors unused where possible. A 48-frame measurement run on a 16-logical-processor machine took 2.93 seconds with four workers and 1.32 seconds with fourteen, with identical star counts (peak working sets 251 and 762 MiB). This measures stellar/background processing, not whole-project speedup. No full-resolution image collection or stack is held in RAM. Windows uses a verified process ceiling based on physical RAM minus its reserve, rounded to an allocation boundary. Fixed 6 GiB process, 3 GiB heap and 256 MiB frame caps have been removed. Per-frame estimates include a 64 MiB working margin and are limited to half the machine budget. Both image passes, previews and star detail acquire atomic memory reservations, sampling physical headroom for every frame. The OS/desktop reserve remains one quarter of RAM up to 8 GiB; page-file capacity does not satisfy it. Sustained external pressure stops safely after active frames finish and saves completed measurements.
Small JSON metadata stores decisions, notes, evidence and compact masks. Binary versioned caches hold stars, sky maps and forced photometry, validated against source size and modification time and reference identity. A new engine version invalidates analysis without losing human labels. Repeated analysis uses cached measurements and forced apertures when valid. Project writes replace a temporary file atomically. Cancellation leaves analysis incomplete and export disabled.

View file

@ -0,0 +1,9 @@
# AstroFrame 1.1.0
- Processing uses more available RAM and CPU capacity. Fixed memory caps have been replaced with a physical-RAM budget and atomic reservations for concurrent frames, retaining headroom for the operating system and other applications. Processing pauses under RAM pressure and preserves completed work if pressure persists.
- Scroll over a preview to zoom around the cursor and drag with the left mouse button to pan. Zoom and position carry across frames and the nearby comparison. **Reset view** fits and centers both previews. Overlays follow the transformed image.
- Ordinary launches begin with a fresh, empty session. Saved projects and decisions remain on disk; explicit project opening and update restarts still restore the requested project.
- **Settings · Update available** turns cerulean blue when an update is available.
- The native startup background matches the application's charcoal grey.
Validation covers engine/resource policy, updater safety, mocked publication, routed mouse events, preview resets and frame switching, rendered layouts at normal and minimum size including 144 DPI, and Windows native window behavior. A 48-frame measurement benchmark on a 16-thread machine improved from 2.93 seconds with four workers to 1.32 seconds with fourteen workers, with identical star counts; this is not a whole-project speed guarantee. Development verification uses workspace fixtures and a fake updater, leaving installed copies untouched.