pick: hit-test the mark a symbol draws, at the size the view shows it - #43
Merged
Conversation
The cursor pick tested a 96 tile-unit radius around a symbol's anchor. A buoy or a beacon draws nearly all of its mark ABOVE the charted position, so the mariner clicked the topmark that is visible and the pick missed the aid. The report then fell through to the areas underneath. QuerySurface now also tests the symbol's drawn box. symbols.bounds gives the painted extent about the pivot, asymmetric and with the stroke half-width included; QuerySurface.inSymbolExtent applies pushSymbol's transform inverted, so the query point is compared in the symbol's own upright frame. The two callers in chart.zig hand the surface the day symbol store, which is enough because a palette supplies only fill and stroke colours and not geometry. Units: pushSymbol draws at k = scale * 100 * dev with dev carrying px_per_tile / 256, and tile units convert to canvas px at px_per_tile / EXTENT. The px_per_tile cancels, so a symbol's extent in tile units is scale * 100 * EXTENT / 256, exact at any zoom. That makes one reference pixel worth EXTENT / 256 = 16 tile units, which confirms the existing radius of 96 as the 6 px it claims to be. The box is UNIONED with that radius, so the radius stays a floor and this can only add hits. A null store, an unknown symbol name, or a symbol with no geometry all fall back to the radius alone. Areas and lines are untouched. A sounding and a label keep the anchor test. A sounding's digits sit around the anchor by their baked pivots and stay inside the radius, and reproducing either extent needs state the query path does not carry. Checked against US5MD1MC at the reported click, 7.95 points above Chesapeake Harbor Entrance Light 2: the pick reported 5 features and no aid before, and now reports BCNLAT and DAYMAR. The pick at the anchor is unchanged, feature for feature.
The query replays the tile at a rounded, clamped zoom, so the view is rarely at
that tile's own level. A tile shown above its level is stretched, and both terms
of the pick were measured as if it were not.
A symbol's drawn box used a fixed 4096/256 tile units per pixel. The box was
therefore 2^(view_zoom - tile_z) times too large, without bound once the view
runs past the archive's deepest zoom. The anchor radius held a fixed 96 tile
units and grew the same way, against a comment that promised a constant
on-screen distance.
unitsPerPx returns the ratio at the view zoom. Both terms use it. At a tile's
own level the radius is still 96 units, so a pick there does not change.
Measured on US5MD1MC (archive z0..16), the deepest click above a beacon's anchor
that still reports it:
view zoom before after
16 12 px 12 px
17 24 px 12 px
18 48 px 12 px
19 96 px 12 px
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A pick tested a ~6 point radius around a symbol's anchor. Buoys and beacons draw almost their whole mark above the anchor, so clicking the topmark you can see missed the aid entirely and the report fell through to the areas underneath. Clicking Chesapeake Harbor Entrance Light 2 on its topmark, 7.95 points above the anchor, returned four features ending in M_NPUB and no aid.
QuerySurface now carries the symbol store and tests the box the symbol draws, unioned with the anchor radius so the radius stays a floor and the change can only add hits. At the anchor itself, before and after report the same 19 features in the same order.
The second commit fixes the units. The query replays the tile at a rounded, clamped zoom, and a tile shown above its level is stretched. Both the symbol box and the anchor radius were measured as if it were not. unitsPerPx returns the ratio at the view zoom; at a tile's own level it still yields 96 units, so picks there do not change.