a-b-street/abstreet
1
1
Fork 0
mirror of https://github.com/a-b-street/abstreet.git synced 2024-12-20 04:42:16 +03:00
Code Issues Projects Releases Wiki Activity
6,832 Commits 85 Branches 166 Tags 395 MiB
c3e35a9a3f
Commit Graph

6832 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Dustin Carlino
6c1c6a3235 Fix crash when changing speed limit. At some point, we started changing lane IDs for every road edit and never updated this. Closes #685 2021-07-02 11:54:59 -07:00
Dustin Carlino
4fd86c8a41 Entering edit mode is slow on large maps, because clearing the simulation recreates parking state for every building. Parallelize it to speed it up at least a bit. 2021-07-02 11:47:05 -07:00
Dustin Carlino
866ecd5f88 Fix a bug in efaf6085e4 that was using walking paths for vehicles. Now prebaked data doesn't budge. 
And add some more timing to demonstrate why editing a single lane in a
huge map is so slow...
 2021-07-02 11:25:02 -07:00
Dustin Carlino
f4c8a508da Keep a JSON file with a summary of prebaked simulations, to more easily detect regressions with broken trips. 2021-07-02 10:07:37 -07:00
Dustin Carlino
1b8d05b264 Upload data for Brooklyn. Also grabbed fresh NYC OSM data for the Manhattan maps too 2021-07-01 13:11:14 -07:00
Trevor Nederlof
3faf9893a9 Add downtown brooklyn as a nyc map 2021-07-01 12:27:58 -07:00
Dustin Carlino
9f4fac58a7 Workaround a crash trying to render building 4582 in Green Lake, which 
has a short driveway that's crashing make_polygons.

This is more fallout from the f64 serialization change. I don't
understand the root cause yet...
 2021-07-01 10:43:42 -07:00
Dustin Carlino
4f00c8d2bb Phinney runs without gridlock again, so re-enable it. Pretty sure 
f57278e9cd fixed it.
 2021-06-30 16:51:03 -07:00
Dustin Carlino
efaf6085e4 Prepare for exiting driveways on the off-side by switching to 
specialized PathRequest constructors
 2021-06-30 12:35:22 -07:00
Dustin Carlino
137e9a0c2c Fix up some places affected by the f64 serialization: 
- offstreet_parking_length in importer config
- backwards compatibility for map edits
- fixing up the baked-in proposals
- working around a few PolyLine bugs that happen with the new rounding
- Don't regenerate actdev scenarios yet -- the upstream JSON format is
  out of date, will have to fix separately.
 2021-06-30 06:35:42 -07:00
Dustin Carlino
057ac5d020 Serialize trimmed f64's in the geom crate as i32, since the f64's are 
trimmed to 4 decimal places anyway. Do this for Pt2D, Distance,
Duration, Speed, Time.

Regenerating everything...
 2021-06-30 06:35:42 -07:00
Dustin Carlino
70a9508415 Change the CityPicker to merge the list of maps and cities from the local filesystem and the manifest. 
This fixes various problems when importing local maps (in new or
existing cities) and getting out of sync with the manifest.
 2021-06-25 19:37:42 -07:00
Dustin Carlino
126c62d6ba Simplify CityPicker logic: just use the optional city.bin to draw districts. Always use the filesystem to create the buttons. 2021-06-25 19:11:37 -07:00
Dustin Carlino
8546eb4b5f Make the timer hierarchy for --regen_all or any city with many maps more easy to understand 2021-06-25 15:29:28 -07:00
Dustin Carlino
3cda1342ab Fix nondeterminism in z-ordering for lanes, hopefully removing noise from the screenshot diff test. 
(Ideally there'd be no overlapping lanes in the first place, but that's
quite hard)
 2021-06-25 12:29:55 -07:00
Dustin Carlino
5161474955 Stop building huge_seattle twice in the importer workflow 
I have some fuzzy vision of declaring all importer pieces as a dependency graph, with optional caching of some input, that would express this kind of thing better. But... in the meantime.

Now regenerating everything...
 2021-06-25 12:18:04 -07:00
Dustin Carlino
f57278e9cd Enable the new walking turn algorithm for consolidated intersections 
only. It has much simpler logic and produces correct results for these
intersections, but it has degradations in some cases -- so limit where
we use it.

But in the meantime, this makes realistic crosswalks over Aurora,
fixing crazy walking detours there.

Not regenerating yet.
 2021-06-24 12:41:38 -07:00
Dustin Carlino
c0ce76a92a Two slight adjustments to speed up the map importer development 
workflow.
 2021-06-24 12:13:47 -07:00
Dustin Carlino
44171500b7 Get the Green Lake map to complete without gridlock!!! 
- Allow blocking the box around two complex intersections in Green Lake.
  This makes the vehicle behavior much more realistic there, by visual
  inspection.
- Amp up offstreet parking to 10 per building. I noticed the simulation
  completes easily with --infinite_parking. This is an approximation of
  that. We make really bad guesses about carpooling and the amount of
  parking available around here, so effectively just remove it from
  consideration for now.
 2021-06-23 13:56:42 -07:00
Dustin Carlino
1bfcba6165 Avoid crash: 
1) Change the language in a map with lot of options
2) Switch to a map with less language options
3) Open settings panel, trying to create dropdown
 2021-06-23 13:13:43 -07:00
Dustin Carlino
8c2b84c37c Keep the trip info panel open when the user clicks before/after and has to initially load the unedited map. #684 2021-06-23 12:52:18 -07:00
Dustin Carlino
9f22d4893a Fix a crash with time-warping to the past when the map is edited. See #684 for an explanation. 
Verified working on native. It should also be fine on web -- loading the
unedited map has to be a whole separate state because of web IO in the
first place.
 2021-06-23 10:29:10 -07:00
Michael Kirk
321c9d9c40 Label contingency matrix axes 2021-06-23 10:13:51 -07:00
Dustin Carlino
799fdfbbc0 Don't start lane-changing while still being a laggy head. Fixes the crash. #392 
Now prebake data again, and upload everything
 2021-06-22 18:36:35 -07:00
Dustin Carlino
190ee4b08f Fix the path v2->v1 and opportunistic lane-changing costs to heavily penalize cars from using bus lanes. 
This doesn't avoid the aforementioned lane-changing crash, but by manual
inspection, it's keeping cars out of bus lanes.
 2021-06-22 17:47:05 -07:00
Dustin Carlino
48076a4c5b Better stop sign placement -- degenerate roads shouldn't get them. Not 
sure why this wasn't done before.

Regenerated all maps... but this happens to expose some lane-changing crashes.
 2021-06-22 17:05:18 -07:00
Dustin Carlino
e08ddba241 Bring in fresh Seattle OSM data with the new cycletrack mapped around 
Green Lake, and fix some missing support for bidi cycletracks.
 2021-06-22 08:31:16 -07:00
Dustin Carlino
678ed84e4c Fix a broken Dropbox link in the importer pipeline. The PSRC raw data is now just in S3. 2021-06-21 21:41:05 -07:00
Dustin Carlino
69152f00ad Fix critical dynamic blockage math bug. Stops some crashes in some cases, including the tests -- which run again. Manually verified the turn goldenfile diff in the UI and regenerated prebaked data. #392 2021-06-21 20:37:11 -07:00
Dustin Carlino
8db88d0890 new release 2021-06-20 20:08:06 -07:00
Dustin Carlino
0f0639d67c Tweak front page docs a bit, [rebuild] [release] 2021-06-20 18:34:02 -07:00
Dustin Carlino
03db0675cc Add a Seattle map covering Green Lake 2021-06-20 10:51:30 -07:00
Dustin Carlino
786bb1159d Fix the lane widths appearing the same in the dropdown due to rounding. #597 2021-06-20 10:30:07 -07:00
Dustin Carlino
ae3d4e2363 Fix loading edits that change the number of lanes in a road. We need to compare the old version of the road in the edits with the basemap, NOT the edited version of the road. #597 2021-06-20 10:06:38 -07:00
Dustin Carlino
d8cdfaba05 Make the multi-road editor only check and edit lanes_ltr, not speed limit. The Aurora cycletrack edit stopped matching because the speed limit happens to change along Aurora. We can think through a better UX to handle both, but I suspect most use cases will focus on lane configuration. #597 2021-06-20 10:00:05 -07:00
Dustin Carlino
e6e6fb267d Reset current selection after editing multiple roads. This sometimes crashes if the selected lane gets deleted or assigned a new ID. Fixes a crash trying to make the West Green Lake cycletrack. #597 2021-06-20 09:50:33 -07:00
Dustin Carlino
3b748ee9fa Implement dynamic blockages and use them for lane-changing. #392 
This is conceptually simple and improves the behavior of vehicles
that're following a LCing leader.

Regenerate prebaked results -- no new gridlock!
 2021-06-19 12:35:19 -07:00
Dustin Carlino
1e43d929e5 Implement dynamic lane-changing in a simple way: 
- only triggered when a vehicle becomes Queued
- Only one adjacent lane, no contraflow (crossing the road's center
  line)
- Don't return to the original lane after passing
- Using a static blockage in the old lane (so other vehicles will wait
  too much)
- Only using the new lane to determine position (so visually a car will
  clip a bike as they pass)

Haven't regenerated prebaked data. #382
 2021-06-19 12:35:19 -07:00
Dustin Carlino
0bcf01fd05 Avoid crashing in recalculate_intersection_polygon due to polylines not 
containing some points. Happens at i524 in Arboretum and in the West
Seattle proposal.

I don't understand yet why this is possible in the first place, but
crashing is pretty bad in the meantime.
 2021-06-19 11:29:01 -07:00
Dustin Carlino
8f199177f5 Upgrade to Rust 1.53 
- arrays are now iterable directly
- switch to using BTree{Set,Map}::retain!
- a round of clippy
- regenerate scenarios and prebaked data; not sure why, but there's a
  diff
 2021-06-17 23:04:35 -07:00
Dustin Carlino
0f7299f091 Found the bug with the previous commit. Keeping this separate to point out the issue. Reasoning using get_active_cars is super dangerous; likely going to remove it. #555 
But finally, regenerate prebaked data successfully and commit all of
this. It's done. Blocking driveways on exit works now
 2021-06-17 13:39:26 -07:00
Dustin Carlino
841336ad15 A grand finale: refactor logic to update follower vehicles, and call that new logic when clearing static blockages. #555 
And previous commit was wrong about Unparking needing distances; it
needs distances OF ANOTHER QUEUE.
 2021-06-17 13:01:00 -07:00
Dustin Carlino
1528b954c6 Carefully change the Unparking and IdlingAtStop states to require 
distances... #555

Prebaked actually broke with the previous commit in Arboretum, because a
car waits behind a blockage, then suddenly warps ahead. But now we can
fix this...
 2021-06-17 12:18:25 -07:00
Dustin Carlino
1dc973c7fa And finally: enforce static blockages while a vehicle exits a driveway onto a far lane! #555 
Vehicles now wait behind blockages, but one more little bug... after the
driveway is clear, the waiting vehicle jumps forward. Need to reset
their crossing state, same as waiting behind a bus.
 2021-06-17 09:27:48 -07:00
Dustin Carlino
f5dc28900b The big rip: change Queue to store more than just actively moving vehicles. Expose these in the API and update a few callers. #555 
Still no behavioral change -- not inserting blockages into the queue
yet, just representing them. Verifying no prebaked changes still.
 2021-06-17 08:51:07 -07:00
Dustin Carlino
ab539f1efd Creeping towards #555: make the queue of cars in Queue private, and start some proper APIs for managing it. We need to change this to model blockages, and this is a small step towards that. 
No behavioral change; verified by regenerating prebaked data
 2021-06-16 21:35:57 -07:00
Rémi Lauzier
1a4b997818 Fix some clippy warnings 2021-06-16 16:21:35 -07:00
Dustin Carlino
cbce654800 Prepare to have vehicles exiting driveways block any intermediate lanes. #555 
This is just wiring together all of the state transitions. Haven't
decided how to internally implement this in Queue yet.
 2021-06-16 16:02:17 -07:00
Rémi Lauzier
dd050ea425 Update some dependency 2021-06-16 15:51:03 -07:00
Dustin Carlino
321e7cd0ee Allow exiting a driveway onto any lane (in the same road direction). #555 
- Support this at the pathfinding level, when transforming v2->v1
- Adjust how the vehicle's body is rendered as it exits a driveway onto
  a farther lane

No support yet for blocking any intermediate lanes; vehicles may clip
through each other without any conflict. Planning to add that
separately.

Regenerating all scenarios and prebaked data...
 2021-06-16 15:46:09 -07:00