🚗 iOS app for finding public parking lots

Ich habe beispielhaft für die Centrum-Galerie für das Jahr 2015 eine Vorhersage der Belegung (in %) mit Machine Learning entwickelt. Das könnte man als Experimental-Feature in die App einbauen.

Ich stelle mir so eine Art Slider vor, wenn man auf die Centrum-Galerie geht, wo man |======+30min=======in 1h=========in 2h=========in 3h==.. und auch am 18.08.2015 um 15Uhr über ein Date-Picker Tool in Android eingeben kann und dann die Belegung von der App zurück bekommt. Dazu müsste im einfachsten Fall eine .CSV geparsed werden und der korrekte Wert daraus entnommen und angezeigt.

funktionsmäßig klappt die Städte-Auswahl bei mir nicht wirklich gut, wenn ich die Stadt zum 1. Mal lade. Ich wähle die Stadt, gehe auf "Fertig", warte 1sec, dann kommt irgendein schwarzer Balken oben rein, der einen Text zeigt und geht nach gefühlt 10ms wieder weg. Dann muss ich nochmal auf "Fertig" drücken und komme dann erst zurück. Klappte aber irgendwie nur zu Beginn nicht so richtig - mittlerweile klappt es. Weiß nicht ob du Prefetchst und daher nur die ersten paar Sek nach dem App-Start die Städte noch nicht geladen sind, aber den Use Case dass ein Benutzer sofort nach dem Start die Stadt wechselt, kannst du ja nochmal testen.

Just gonna quote @sibbl here :P

Formula should be: sqrt(Entfernung^2 + (1/((2*(1-Belegung))^2 + 1))*Belegung^2) und dabei Entfernung in Meter und Belegung normiert 0...1

1/((2*(1-Belegung) looks like this:

reason is, that crowded parking lots are rated down.

As a user from Dresden i would expect the default city after starting the app to be Dresden. Instead the selected city was Oldenburg. If the users position can't be used, one could add a dialog to select a default city.

Just to keep track (checked means added to POEditor):

• [x] euklidian sorting
• [x] unknown server error
• [x] no coords error
• [x] city options
• [x] outdated data warning

This app should be prettier. Everything about it. I've got some ideas...

And also a new icon. Definitely a new icon!

Just gonna collect a few links here: http://flatuicolors.com https://github.com/KrauseFx/TSMessages

Because why not. I have no clue how hard it might be to implement this though. And hopefully it doesn't interfere with the UIRefreshControl.

See this: http://www.raywenderlich.com/76519/add-table-view-search-swift

Stupid issue, just want to collect some minor thoughts from this morning as a checklist:

• [x] Switching a city should result in a refresh of the LotlistVC
• last updated time
  • [x] Add "As of" or "Stand" to the last updated cell
  • [x] Localize that
  • [x] Date of last updated cell should be converted from UTC
  • [x] Warn user on old data. Either if last updated and last downloaded differ by too much or last downloaded is far in the past
  • [x] Tapping on last updated cell should open data source in Safari
  • [x] Change placeholder for last updated cell
• [x] Keep option to not show lots without data?
• [ ] Figure out if app can crash if a selected city is removed from the API

Habe dieser Tage mal einen Reality-Check in der Stadt gemacht: Die freien Parkplätze, die z.B. an der Anzeigetafel auf dem Dr.-Külz-Ring (auf Google StreetView ist da noch keine) angezeigt werden, stimmen nicht überein mit denen, die in der App gerade angezeigt wurden. Waren so +/- 20 Parkplätze Unterschied.

Das ist aber kein Fehler unsererseits, sondern die Tafeln in der Stadt werden nicht konsistent von der gleichen Datenquelle gespeist. Wenn man z.B. um die Ecke herum fährt und zur Centrum-Galerie abbiegt in die Reitbahnstr., dann steht dort wieder eine Tafel und diese Zahlen stimmten exakt mit den Zahlen in der App überein.

Falls also Beschwerden von jemandem kommen sollten: Da können wir nix für! :-)

If the sorting mode is set to distance the table's data is reset to default sorting until a location has been determined. This can be a few seconds and is visually unpleasing to say the least.

Best idea would probably be to remember the distance sorted list temporarily and set it to that before resorting after the real location has been found. The labels can of course remain at waiting for location for this waiting time.

Instead of having three distinct colors for the main states (green, yellow and red), it would be a better idea to incorporate the percentage load into this somehow. So either a filled circle or a bar or something.

I haven't really put a lot of time into this app lately and when I did get around to it, there's just too much to run into, rip out and start over with. Most of that is due to the fact that this was my first iOS project and started out in Swift 1.0 (!). The overall design (code-wise and the UI) really isn't that well thought-out and can be massively improved. I'm getting a strong feeling that the sane thing to do at this point is to just scrap the old project and start over with the lessons learned. Rethink what kind of data the app shows and streamline that.

Some random thoughts on this topic that have been floating around in my head, not an exhaustive list:

• Map view (or split map/table) should be the default
  • What about a drawer like design like the new Maps.app? Does that make sense here?
• Map annotations should display more information, see Windows version of ParkenDD
  • In fact, the Windows UI has always been pretty damn awesome! Cheers, @sibbl 🎉
• It should be much more obvious that more cities besides Dresden exist
  • I'm still all for renaming the application, but all ideas like "Parking", "Parken", "OpenPark" stink 😕
• Navigation to lots should be possible, but it would probably make sense to leave this to other apps (see Karte)
• Keep the codebase clean and use ParkKit
• Throw a Patreon-esque IAP option somewhere, see "old" Overcast or ChatSecure
  • No essential features (if any) should be behind the IAP though
• Evaluate the use of a local trained neural net for forecast info (🏓 @absolute-heike about this)
• Possibly send out push notifications for new cities (would obviously require this to be added to the backend as well)
• Try and make a lot more use of testing and automation tools like fastlane to make publishing updates on this project not so much of a chore^^
• I planned to use more system extensions points quite a way back now, but never got around to that. This app could work pretty well with a Today extension and Watch companion app.
• Evaluate the use of MapBox instead of Apple's MapKit

Throw in some very simple stat tracking sent to @jklmnn's piwik instance. Basically most of what was previously gathered through Fabric:

• which city is being loaded (might as well only be tracking with the API)
• forecast stats for which lot are being viewed (see above)
• what sorting settings are being used
• are experimental cities being displayed

Obviously waiting on #160.