opentripplanner / otp-react-redux Goto Github PK

Currently an issue in leaflet prevents marker images from loading correctly. Temp fix made in 7bc89d4 references node_modules/leaflet/dist/images/ path, but this is not a solution for a production build.

What happens when a mode supplied here doesn't exist in modes available in config? It looks like GONDOLA cannot be removed from the list. Perhaps the modes-panel should filter out any modes that are not available to prevent a mode from sneaking in and messing up plan request results.

Presently, if a GTFS-rt alert has a body value, it is not rendered in the UI. TriMet alerts do not have bodies, so the original UI was designed without a way to render this. Many other agencies do have alert bodies, so we need a way to display these.

Check out the difference between https://newplanner.trimet.org and https://trimet-mod-dev.ibi-transit.com/ . This is an issue with otp-ui or its implementation in otp-react-redux.

https://trimet-mod-dev.ibi-transit.com

https://trimet-mod-dev.ibi-transit.com/

Originally posted by @landonreed in #165

Clicking the Print button in PrintLayout does nothing, at least on Firefox for Android and Samsung Internet (Chromium-based) for Android. It seems to trigger and print prompt on Chrome for Android. You can still print using the browser menu.

This may or may not be related to #133.

There is an alignment issue for late morning/evening times:

Originally posted by @binh-dam-ibigroup in #165

This issues originates from this suggestion to add a config parameter to display narrative distances in metric units. Currently, distances are hard-coded to be shown as imperial.

This issue depends on otpui/#116.

I finally tracked down the root cause of the browser compatibility issues (or at least one of them). hash-obj is not transpiled for IE, so we should not be using it in otp-rr. There may have also been some issues with incorrect polyfills being included, but that should be resolved with the latest version of trimet-mod-otp. Thoughts on an alternative for this package, @evansiroky?

Branch:Dev
Geocoder: nominatim
We are getting the error see attach, when click Plan Trip. In the same time desktop version working without issues.

Appreciate for any advce.

Trip details time format is not consistent with configured time format.

On certain itineraries with large elevation changes, the labels get all jumbled together. See example:

After merging #147, the bus route label in LineItinerary
is shifted as shown below. The correct alignment is the original alignment.

Below is a list of pending issues to resolve. Please edit the issue body directly to add any additional ones identified in PRs but not yet resolved.

• please better style transit route numbers #165 (comment) (route numbers removed for now 54b1c4e)
• please make itinerary cost factors and itinerary sort factors configurable #165 (comment) and #165 (comment) (added in 79c835d)
• Have the settings and timing buttons do something in the batch-routing-panel file. #165 (comment)
• Make save trip actually work in narrative-itineraries #165 (comment) (a68c3df)
• Make addition/update/deletion of intermediate locations change map bounds #165 (comment) (5d2c55e)
• Await fix for intermediate place parsing #165 (comment) (45d92ef)
• other intermediate places enhancements
  • ability to drag the intermediate place map markers
  • ability to drag/drop the intermediate places to reorder them in the form
  • the reverse from/to button should probably just be removed if there is at least one intermediate place

Possibly use woonerf fetch to standardize API calls. Possible concerns: library bloat, relying on Conveyal-specific lib.

When displaying an itinerary with alerts, the expanded alert does not render the alert URL. This prevents directing users of the trip planner to more information about the alert that may be contained on the agency website, which is especially useful if it contains dynamic or detailed instructions.

Current alert rendering

Alert data

Instead, we should add the URL with the hyperlinked More text

"We should phase out the existing state.otp.user portion of the otp reducer [...]"
and move the corresponding code to the new user reducer

Originally posted by @binh-dam-ibigroup in
#167 (comment)

Some web services may require custom headers for authenticating requests. We need a way to define these in the geocoder config and pass them along to libraries we're using. Re: conveyal/isomorphic-mapzen-search#46

I tried to use the example app and followed the instructions but I get the following error: TypeError: _toJS is undefined;\

(I'm on Firefox on Linux, if it matters.)

I've added the config.yaml file and it's contents are:

api:
  host: http://localhost
  path: /otp/routers/default
  port: 8080
# Add suggested locations to be shown as options in the form view.
# locations:
#   - id: 'airport'
#     icon: 'plane'
#     name: 'Portland International Airport (PDX)'
#     lat: 45.587579
#     lon: -122.593084
#   - id: 'zoo'
#     icon: building
#     lat: 45.51010444232195
#     lon: -122.71607145667079
#     name: Oregon Zoo, Portland, OR

### The persistence setting is used to enable the storage of places (home, work),
### recent searches/places, user overrides, and favorite stops. Currently the
### only strategy is localStorage (which is used by default). It also must be
### enabled to show the stored locations (see above).
### TODO: add another server-based strategy
persistence:
  enabled: true
  strategy: localStorage

map:
  initLat: 45.52
  initLon: -122.682
  baseLayers:
    - name: Streets
      url: //cartodb-basemaps-{s}.global.ssl.fastly.net/light_all/{z}/{x}/{y}@2x.png
      subdomains: 'abcd'
      attribution: 'Map tiles: &copy; <a href="http://www.openstreetmap.org/copyright">OpenStreetMap</a>, &copy; <a href="https://carto.com/attributions">CARTO</a>'
      maxZoom: 20
    - name: Stamen Toner Lite
      url: http://tile.stamen.com/toner-lite/{z}/{x}/{y}.png
      attribution: 'Map tiles by <a href="http://stamen.com">Stamen Design</a>, under <a href="http://creativecommons.org/licenses/by/3.0">CC BY 3.0</a>. Data by <a href="http://openstreetmap.org">OpenStreetMap</a>, under <a href="http://www.openstreetmap.org/copyright">ODbL</a>.'

geocoder:
  boundary:
    rect:
      minLon: -123.2034
      maxLon: -122.135
      minLat: 45.273
      maxLat: 45.7445
  MAPZEN_KEY: gAZxU6e1RamAot9wPp0_lA
  focusPoint:
    lat: 45.5223452
    lon: -122.6821804
  # Optional custom Pelias instance (defaults to Mapzen Search)
  # baseUrl: https://geocoder.example.com/pelias/v1

# Use this mode config for the enhanced Transit+ config
modes:
  transitModes:
    - mode: BUS
      label: Bus
    - mode: TRAM
      label: MAX & Streetcar
    - mode: RAIL
      label: WES
    - mode: GONDOLA
      label: Aerial Tram

  accessModes:
    - mode: BICYCLE
      label: Transit + Bike

  bicycleModes:
    - mode: BICYCLE
      label: Own Bike
      iconWidth: 18


routingTypes:
  - key: ITINERARY
    text: Exact Time

### Use this config for the standard mode selector
# modeGroups:
#   - name: Transit
#     modes:
#       - TRAM
#       - BUS
#   - name: Walk/Bike
#     modes:
#       - WALK
#       - BICYCLE

What am I doing wrong?

When a leg diagram has been toggled on, the diagram previews for all legs are highlighted/turned on. The leg-diagram-preview.js component needs to be modified so that it checks that the active leg diagram is the same as the leg provided in its props.

Add routes to keep track of state of application.

• plan query
• mapState (lat/lon/zoom, layers, etc.)
• active itinerary/leg/step

OTP has two different response types in the transit index for routes: long and short.

The short response (which the route type comparator in otp-rr) currently handles contains a Route@mode field, which maps to the OTP transit mode enum values (e.g., BUS and RAIL). However, in the long response from OTP, the mode field is dropped in favor of route_type, which uses the int-based route types from GTFS. So, when a route is clicked in the otp-rr route viewer, the route object is replaced with the long response, and it drops the mode field. Originally, I thought the route was just disappearing from the list, but in fact it was dropping to the bottom of the list (in accordance with the route type comparator logic -- default to 9999).

Similar to this issue the labels in the leaflet map are much too small and barely legible on many devices:

Some transit agencies have simple forms on their website that connect with Google Transit/Maps by sending the place names, time, and date. otp-react-redux should accommodate the same kind of thing, by simply attempting to geocode the incoming place names when parsing the query params.

Add routes to keep track of state of application.

• plan query
• mapState (lat/lon/zoom, layers, etc.)
• active itinerary/leg/step

Would the OTP team support a pull request to replace mastarm with webpack? Webpack is the standard and mastarm is implemented by Conveyal, which appears it's main feature to support AWS which doesn't apply to some OTP installations. Also, webpack is more universal and easier to customize. Perhaps mastarm2 could become a decorator around webpack to preserve the AWS features.

Hello,
I will develop to allow the application to be multi language.

My question is whether this is raised in any way in the project?
How should I develop it following a guideline or using a especific library?
What library do you recommend to make the multilanguage app?
Is there any thread where this is talked about?

Thanks!

Validate the query parameters being sent to an action and also getting received by a reducer.

Changes made for the TriMet implementation added a Plan Trip button in the DefaultMainPanel component. However, this does not appear in the example.js file that is used on yarn start, so there is no way to test the UI or whether any components modified in recent otp-ui PRs are actually working.

As currently designed, otp-react-redux will make two requests to the OTP server for a trip plan: one with realtime updates applied and one with them ignored. However, it appears to set ignoreRealtimeUpdates to true in the query parameters so that any trip that is reloaded from URL (or maybe even after any query parameter is changed after a trip has been planned) will have the updates ignored for BOTH requests.

It'd be nice to include the github commits that the UI was built with somewhere in the UI.

The sorting used for routes displayed in the route viewer does not follow any intelligible pattern. This is due to the fact that when OTP does not find route_sort_order for routes, it defaults to -999.
OneBusAway/onebusaway-gtfs-modules@8c54d57

When loading the OTP web app in a browser, there is a moment where the original page title is visible before it is overridden in responsive webapp:

This was observed in the 511NY trip planner using the trimet-mod-otp library (see https://dz88fe1xr6gid.cloudfront.net/#/, where page title flashes a non-NY title before changing).

It appears that we need to update the page title in the constructor method as well.

The ArcGis geocoder adapter does not correctly handle getting the location from a search query. The response for this query type is simply a GeoJSON-style feature with geometry and properties (not the magicKey/text result that comes back from autocomplete). Currently ArcGISGeocoder#getLocationFromGeocodedFeature only handles the autocomplete style result:

See discussion on #9 for motivation for issue.

In the current design of the route viewer as of 26b9a8a, the route polyline data is fetched anew each time a user clicks on a new route. In order to cut back on the number of API fetches for users that click back and forth between routes, we could store the response for certain routes in a in-browser cache and return that data to reduce the need for API requests.

The exceptions below are produced when trying to open the route viewer from the trip planning view.
No routes are displayed as a result and the entire screen becomes blank.

Fragment where the exception occurs: d=n.longName.split(c)

For feeds with large numbers of routes, especially when multiple feeds/agencies are present, we should provide controls to filter on route_type and agency.

After further team discussion on this:

• The "filters" will just be standard dropdown selectors allowing the selection of either All Modes/Agencies or a single mode/agency.
• For styling, we'll just use the dropdown selector from the otp-ui form (currently used for setting things like max walk distance)
  

Desktop view

Mobile view

Note: this issue is only detectable on a mobile device (using Chrome dev tools to simulate mobile on a desktop is not enough to trigger).

When a user sets the from and to locations via a geocoded result in the location search component, there is a fitBounds invocation in BaseMap that causes an error to be thrown resulting in a blank white screen.

There appears to be an issue with transitive rendering a specific itinerary from the City of Long Beach (NYSDOT).

1. Go to: https://otp-test.511ny.org/#start/40.5885995/-73.6394705/11/City_of_Long_Beach_ea05fbd9-bfc6-413a-ab04-e777b95484dd
2. Plan trip from
   414 Centre St, Long Beach, NY, 11561, USA
   to
   1092 W Beech St, Long Beach, NY, 11561, USA.
3. White screen of death with error:

Uncaught TypeError: this.parent.getName is not a function
    at SegmentLabel.initText (label.js:36)
    at SegmentLabel.getText (label.js:30)
    at SegmentLabel.render (segmentlabel.js:65)
    at eval (default-renderer.js:117)
    at arrayEach (_arrayEach.js:15)
    at forEach (forEach.js:38)
    at DefaultRenderer.render (default-renderer.js:116)
    at Transitive.render (transitive.js:162)
    at TransitiveCanvasOverlay._updateBoundsAndRender (transitive-overlay.js:159)
    at TransitiveCanvasOverlay._updateBoundsAndRender (react-hot-loader.development.js:708)

I'm worried that this block of code to determine the print label will be at risk of diverging from the block in PlaceRow:

Even though they aren't returning the exact same label, there's some shared logic that I have a feeling we would want to co-locate:

Case in point: I think the only reason this pull requests exists is because there is a significant split between the print and narrative itinerary components.

Originally posted by @landonreed in #100

We have a OTP 1.4.0 running and I tried otp-react-redux.

While the OTP frontend itself shows the routes aligned to the streets, the OTP-React-Redux frontend instead displays straight lines.

I believe the information is in itineraries.0.legs.0.legGeometry

How can this be displayed?

Is this a missing feature, a bug or a misconfiguration on my side?

Avoid direct calls to the leaflet map whenever possible and instead send props to the Leaflet map. E.g.:

Once a trip is saved, there is no indication that the trip (itinerary) has been saved. It would be nice to have the itinerary display change accordingly once it is saved.

Originally posted by @evansiroky in #174

Any complex function delaying with displaying time and timezones ought to have tests with all kinds of edge cases. Let's add an issue for this.

Originally posted by @evansiroky in https://github.com/opentripplanner/otp-react-redux/diffs/35

The command to lint this library is only linting some files... not all of them.

Has anyone considered using OTPRR as a PWA? I ran a test site through lighthouse viewer. While there are a few optimizations needed, it could become one (eg. Does not register a service worker that controls page and start_url, font sizes, no meta description).

This would be fantastic bringing full screen, offline (maybe possible), home screen app icons, etc.

Revisit the use of LinkContainer. Earlier attempts have resulted in LinkContainer not being able to get the router info (does it have to do with the React router setup?)

Originally posted by @landonreed in #174