mvglasow

Accessing map data needs to be thread-safe, as this is an operation which happens in all thread candidates: traffic, routing (in the next phase), and in the UI (we still...

csv maps turned out to be challenging, as they support write operations to any map rectangle, and these in turn touch data structures shared by all currently open map rectangles,...

Re items, points to consider regarding thread safety include: * Does the map share items (or parts of their private data structure) between map rects? If not, there is no...

Also, some map implementations have no `item_priv` but tack the `map_rect` to their items instead. Where this is the case and the map rect has no reference to anything shared...

Almost there—map rect and item access should be almost completely thread-safe now, the only thing still missing is items in `mg` maps. Now we just need to take care of...

If the compiler preserves the order of functions in the source code, `#03` is likely `xml_parse_text`, given its offset (just a few bytes after `xml_parse_file`). `#01` and `#02` might be...

Tried reproducing this today under real-world conditions but could not provoke a crash. Even killing Navit the hard way (by long-pressing Back, needs to be enabled in developer settings) and...

Some changes were made (also related to crashes on startup) while adding TraFF 0.8 support, which might have alleviated this issue. I suggest we keep this open for a while,...

I have not observed any further crashes, despite regular usage including some long trips; therefore I’m closing this for now.

I am still experiencing this issue. This time I have saved a logcat and a traffic.xml file at the time of the crash. Boundaries of the issue: * First route...