Skip synthetic "_simple_" line style in catalogue pre-warm (#286)

[philliphoff]

Summary

Fixes the three KeyNotFoundException throws on every S-101 dataset load reported in #286.

The S-100 Part 9A Lua portrayal model emits LineInstruction / AreaInstruction outline references using the inline _simple_ line-style sentinel (from PortrayalModel.lua's SimpleLineStyle). Its colour, width, and dash pattern travel on the instruction itself — it is never a named complex line style in the portrayal catalogue.

CataloguePreWarm resolved it like any other line style, so each load threw:

Line style '_simple_' not found in the portrayal catalogue.

…three times. The exceptions were caught and discarded but never cached, so they were re-thrown on every subsequent load.

Change

• New LineInstruction.SimpleLineStyleReference constant ("_simple_") documenting the Part 9A sentinel.
• CataloguePreWarm now skips it for both LineInstruction.LineStyleReference and AreaInstruction.OutlineStyleReference. Rendering already drew these as plain inline lines (colour token present ⇒ no catalogue lookup), so portrayal output is unchanged — minus the exceptions.

Tests

• CataloguePreWarmTests — unit cover: the sentinel is not resolved against the catalogue; named line styles still are.
• S101SimpleLineStyleRegressionTests — skippable real-data regression: building an S-101 vector portrayal raises zero KeyNotFoundExceptions. Verified against UKHO/IC-ENC trial cells; reverting the fix reproduces exactly the 3 throws from [Bug]: Performance of S101 multiple datasets #286.

Full EncDotNet.S100.Pipelines.Tests suite: 584 passed, 1 skipped.

Out of scope

This PR does not address the per-dataset memory growth (50–100 MB/cell) also noted in #286 — that stems from the render-side caches (Mapsui scene/style/pattern-clip/tile) that back smooth pan/zoom, and warrants a more careful eviction-policy design. Tracked in a separate follow-up issue.

Closes #286 (exception portion); memory portion tracked separately.

[github-actions]

Performance Gate

✅ PASSED — no regressions.

Threshold: 10.0%, MAD multiplier (k): 3.0, retry-zone mult: 2.0×

Scenario summary

Scenario	Status	Δ median (%)	z (Δ/MAD)	Base median (ms)	Samples (b/c)
exchange-set-open	✅ pass	+2.6	+2.15	1.27	5/5
s101-portray-cold	✅ pass	+3.0	+0.19	0.24	5/5
s101-portray-warm	✅ pass	-1.5	-0.07	0.25	5/5
s101-real-cold	✅ pass	+0.1	+0.01	0.27	5/5
s101-real-warm	✅ pass	-5.9	-0.34	0.28	5/5
s101-render-warm	✅ pass	+1.0	+0.06	0.23	5/5
s102-coverage	✅ pass	+0.6	+0.15	0.16	5/5
s102-coverage-open	✅ pass	-4.1	-0.57	3.01	20/20
s102-coverage-render-large	✅ pass	-4.8	-0.39	0.21	5/5
s102-real-warm	✅ pass	-5.5	-0.72	0.31	5/5
s111-real-warm	✅ pass	+5.3	+3.22	0.27	5/5
s124-vector	✅ pass	+5.2	+0.68	0.19	5/5
s201-vector	✅ pass	+1.8	+2.67	0.18	5/5

exchange-set-open

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	1.27	1.30
Baseline MAD (ms)	0.02	—
Δ median	—	+2.6%
z (Δ/MAD)	—	+2.15

Spans (sum of all iterations)

Span	Baseline (ms)	Candidate (ms)	Delta	Status
s100.asset.read	2.49	2.75	+10.3%	❌
s100.exchangeset.parse	45.24	45.54	+0.7%	▫️

Metrics

Metric	Baseline	Candidate	Delta	Status
s100.asset.read.duration	0.21	0.18	-11.5%	✅

s101-portray-cold

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.24	0.24
Baseline MAD (ms)	0.04	—
Δ median	—	+3.0%
z (Δ/MAD)	—	+0.19

s101-portray-warm

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.25	0.24
Baseline MAD (ms)	0.05	—
Δ median	—	-1.5%
z (Δ/MAD)	—	-0.07

s101-real-cold

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.27	0.27
Baseline MAD (ms)	0.03	—
Δ median	—	+0.1%
z (Δ/MAD)	—	+0.01

s101-real-warm

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.28	0.26
Baseline MAD (ms)	0.05	—
Δ median	—	-5.9%
z (Δ/MAD)	—	-0.34

s101-render-warm

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.23	0.23
Baseline MAD (ms)	0.04	—
Δ median	—	+1.0%
z (Δ/MAD)	—	+0.06

s102-coverage

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.16	0.16
Baseline MAD (ms)	0.01	—
Δ median	—	+0.6%
z (Δ/MAD)	—	+0.15

s102-coverage-open

Iteration statistics

Stat	Baseline	Candidate
Samples	20	20
Median (ms)	3.01	2.88
Baseline MAD (ms)	0.21	—
Δ median	—	-4.1%
z (Δ/MAD)	—	-0.57

Spans (sum of all iterations)

Span	Baseline (ms)	Candidate (ms)	Delta	Status
s100.dataset.open	457.07	459.78	+0.6%	▫️
s100.hdf5.dataset.read	166.51	167.94	+0.9%	▫️
s100.hdf5.file.open	20.24	19.36	-4.3%	▫️
s100.hdf5.open	18.74	18.67	-0.4%	▫️

Metrics

Metric	Baseline	Candidate	Delta	Status
s100.hdf5.read.bytes	36456.00	36456.00	+0.0%	▫️
s100.hdf5.read.duration	27.34	27.12	-0.8%	▫️
s100.hdf5.read.duration	32.12	32.56	+1.4%	▫️
s100.hdf5.read.duration	11.22	11.33	+1.1%	▫️

s102-coverage-render-large

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.21	0.20
Baseline MAD (ms)	0.03	—
Δ median	—	-4.8%
z (Δ/MAD)	—	-0.39

s102-real-warm

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.31	0.30
Baseline MAD (ms)	0.02	—
Δ median	—	-5.5%
z (Δ/MAD)	—	-0.72

s111-real-warm

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.27	0.29
Baseline MAD (ms)	0.00	—
Δ median	—	+5.3%
z (Δ/MAD)	—	+3.22

s124-vector

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.19	0.20
Baseline MAD (ms)	0.01	—
Δ median	—	+5.2%
z (Δ/MAD)	—	+0.68

s201-vector

Iteration statistics

Stat	Baseline	Candidate
Samples	5	5
Median (ms)	0.18	0.18
Baseline MAD (ms)	0.00	—
Δ median	—	+1.8%
z (Δ/MAD)	—	+2.67

---

Generated by EncDotNet.S100.PerfReport gate command