discontinous railways #3
Labels
No labels
claude-wip
map-report
map:is
map:se
needs-review
No milestone
No project
No assignees
2 participants
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
art/ismap#3
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Polish railways have sections disappear at z10. Every part I've seen where this happens has more tracks than the sections that remain visible. Question is, rather than rendering the "smaller" tracks, wouldn't it make sense to merge both railways and roads with close parallel lines into one thicker/higher priority one and that would solve this problem?
Location: z14 / 52.43021 / 21.42863
Area: circle r=1268 m @ 52.31349, 21.15892
Location: z11 / 52.22458 / 21.03393
Area: circle r=1181 m @ 52.25280, 21.10094
railwaysat coarse zoom — it is now the biggest thing in a continental tile #9Fixed on
fix/rail-merge-and-pl-classification(commite6abf49, not merged).Your diagnosis found the right place, but the mechanism is different — and simpler.
The sections that vanish at z10 are not being lost because they are small parallel
tracks. They are being lost because they are misclassified. BDOT10k's
FUNKCJATORUwas tested for
%stacyjny%first, sotor szlakowy stacyjny— which is the runningline where it passes through a station — was filed as
siding, and z10 drops sidings.Every main line therefore had a gap at every station.
That is exactly consistent with what you saw: stations are precisely where there are
more parallel tracks, so "sections disappear where there are more tracks" is the same
observation seen from the other side.
The data settles which reading is right.
NUMERLINII, the PKP line number, is what"this is a running line" looks like in the source:
FUNKCJATORUtor szlakowymaintor szlakowy stacyjnymaintor zwykłybranchtor zwykły stacyjnysidingbocznicasidingThe word that classifies a track is
szlakowy(running line), notstacyjny(in astation), so
szlakowyis now tested first and both its forms aremain. This restores5 213 km of running line and takes the average numbered PKP line from 6.16 pieces to
3.79 at z10 (3 901 → 2 783 nationally). Line 21 at the spot you reported goes from 7
fragments with 2 249 m missing to a single continuous 21.6 km piece.
Your parallel-merge idea is in as well, because it is right for a different reason.
It is not what was breaking the lines, but it is what stops a station throat or a
freight yard from being a solid blob of ink, and it is a large byte win — see #9, where
it does most of its work. Implemented as
merge_lines+parallel_merge_pixels+parallel_priorityinimporter/layer.go: a chain lying 95% inside a better-rankedchain's n-pixel corridor is dropped, and "better-ranked" is the class ladder, so the
survivor is the highest class in the corridor — the "merge into one higher-priority
line" you described. It is deliberately generic and reusable for roads, which is
#10; the one thing to solve first is that
ST_Collectgathers a group into a singlegeometry, which is fine for rail but wants a spatial split for a country's road network.
Rendered checks at your location and at Warszawa z11 look right — yards read as legible
track rather than a blob, and the through routes are continuous.
railwaysat coarse zoom — it is now the biggest thing in a continental tile #9Merged to
mainase6abf49and pushed. Closing.