make validator results indicate whether results may change by loading new objects

[aceman]

Some validator warnings only happen when objects are downloaded and some neighbouring objects are not. E.g. this happens with bridge relations (looks like the type of members that are not downloaded can't be checked and issue a warning of wrong type).
Also the checker of highway=*_link tags produces bogus warnings if the roads connected to the one being analysed are not loaded (ticket #9043).

See also ticket #9544.

Is it possible to determine which tests may produce bogus warnings in this scenario? Can we add a hint to the editor that he should try to load neighbouring areas?

[Klumbumbus]

Meanwhile some tests were changed, so they so only test in the downloaded area. Are there any left, which should be updated?

[Klumbumbus]

I found one: a secondary roundabout with two tertiary roads connected and two secondary roads not downloaded gives false positive "incorrect roundabout (highway: secondary instead of tertiary)"

[aceman]

I was thinking e.g. about the "ford should be on a node where water and road cross".

Are the bridge relations I mentioned in the description fixed too?

[Klumbumbus]

ford warning is fixed.
do you have an example of such a bridge relation? I don't really know what you mean. The bridge relation preset has only members with count 0-...

[aceman]

I meant the "under" and "across" members (roads below/on the bridge). If those ways were not loaded in JOSM, there was a warning in the validator. But I also can't reproduce it now. Maybe this was also fixed in another ticket.