Multi download

[westnordost]

No description provided.

[smichel17]

Follow-up question: how large is the stored quest data? Does it (or should it) interact with the "map cache size" setting?

[westnordost]

It doesn't. Quests are not considered part of the map. Map cache is only for the background map. Unsolved quests are

[smichel17]

Unsolved quests are

I think you accidentally

[westnordost]

... deleted automatically after a certain interval

[westnordost]

Hmm, I think I am done.

Anyone want to review the code or give it a go in a test?

I implemented what @smichel17 and @matkoniecz suggested. On my Samsung S10e, download is so fast that after I tapped through the tutorial, all the quests are already there. 5s for initial scan (zoom 16 tile around my location - 360x360m) yielding 140 quests, then another 8 seconds for the download in 600m radius around my location, yielding another 2200 quests.

The download logic works like that: First download the z16 tile which encloses the user's location (around 360x360m), then from that, calculate the quest density. The download strategy when in WiFi targets downloading around 1000 quests around the user, so it calculates the estimated download radius based on the measured quest density and then issues another download. In my case, that was around 600m. The mobile data download strategry aims at having around 500 quests around the user, so downloaded radii will be smaller.

[westnordost]

Interesting classes:

• AVariableRadiusStrategy.kt - class that decides over whether a new download is started automatically or not
• AddRailwayCrossingBarrier.kt - see how crossings with private ways are filtered out programmatically
• AddSummitRegister.kt - see how closeness to hiking route relations is determined programmatically
• OsmApiQuestDownloader.kt - how the quests are downloaded now
• AddCycleway.kt - complex programmatic filtering, involving closeness-checks to other ways
• (more - I didn't look through everything myself yet)

[smichel17]

I can't comment on the parsing without spending more time learning OSM details than I want, but I found one potential issue.

[westnordost]

Thanks for trying it out, @smichel17 !

[westnordost]

• AddHousenumber - complex filter checks programmatically
• AddRecyclingContainerMaterials - complex programmatic filter

[westnordost]

Unfortunately, the problem here is that in small bounding boxes, quite a few buildings will be excluded because of this. If a housenumber quest was downloaded before as part of a big bounding box and then another small download is being executed, unsolved housenumber quests on buildings on the edge will even vanish.

The prior solution was to expand the bounding box to include all the buildings and then download the housenumber nodes based on that enlarged bounding box. With the solution now, this is not possible (or: not expedient) anymore - we have that one bounding box with the data and we should create all the quests based on the downloaded data.