aboutsummaryrefslogtreecommitdiff
path: root/docs/tomes
diff options
context:
space:
mode:
authorValentin Popov <valentin@popov.link>2026-07-18 12:28:27 +0300
committerValentin Popov <valentin@popov.link>2026-07-18 12:28:27 +0300
commit564edf0bc1d5fed1cbfc00db2ba5c4cd9d1b94ea (patch)
tree5774de073f48c44195c149298bb67d4f2f12da95 /docs/tomes
parent2474e480200a8dfa260a069730c19c4bb2cbdf8e (diff)
downloadfparkan-564edf0bc1d5fed1cbfc00db2ba5c4cd9d1b94ea.tar.xz
fparkan-564edf0bc1d5fed1cbfc00db2ba5c4cd9d1b94ea.zip
feat(runtime): trace visual cache sizes
Diffstat (limited to 'docs/tomes')
-rw-r--r--docs/tomes/05-render.md8
1 files changed, 8 insertions, 0 deletions
diff --git a/docs/tomes/05-render.md b/docs/tomes/05-render.md
index d390a9a..cf51b41 100644
--- a/docs/tomes/05-render.md
+++ b/docs/tomes/05-render.md
@@ -1133,6 +1133,14 @@ probe recorded MAT0 request 3 at 52 nodes / 51 edges and request 4 at 55 nodes
they do not change traversal order, node or edge identities, provenance,
validation, cache semantics, or rendering.
+The same throttled snapshots now retain the number of distinct validation-cache
+keys for WEAR, MAT0, and TEXM before the request. In a fresh 50-second Part 2
+probe, MAT0 request 3 recorded cache counts 2 / 3 / 3 and request 4 recorded
+3 / 4 / 4, while the graph moved from 52 / 51 to 55 / 54 nodes / edges. The
+counts distinguish cache growth from graph growth without adding per-request
+file writes or asserting which parser, archive I/O operation, or allocation
+caused the elapsed interval.
+
`fparkan-game --load-progress` now initializes its file with `Starting` and
appends each later mission-load event instead of replacing the prior line. A
bounded probe can therefore retain both MAT0 and TEXM request milestones even