Changes between Version 1 and Version 2 of Ticket #12427, comment 24


Ignore:
Timestamp:
2016-01-25T16:00:42+01:00 (8 years ago)
Author:
cmuelle8

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #12427, comment 24

    v1 v2  
    55So yes, in theory the exact case is solved, but ways to balance between those expensive methods and ''good'' approximations that reduce costs in practical applications I find to be sparse.  We have already "highlight" on objects you hover the mouse over, if we find methods of inexpensive, good approximations we can enable distance and area calculation for each object hovered over (ideally caching the result, like you cache styles, or like the multipolygon-cache caches areas)..
    66
    7 The search is (in the very long run), one for iterative/recursive methods that can be interrupted at any time, such that the calculation is not lost but gives a usable approximation. E.g. for the area calculations it'd be nice to have an algorithm in the end that starts on all segments at the same time and if interrupted early, outputs the best approximation that could be made in the time it was permitted to run.  Of course, it'd be nice to now (and eventually deliver that to the user as well) the error class the result is in..
     7The search is (in the very long run), one for iterative/recursive methods that can be interrupted at any time, such that the calculation is not lost but gives a usable approximation. E.g. for the area calculations it'd be nice to have an algorithm in the end that starts on all segments at the same time and if interrupted early, outputs the best approximation that could be made in the time it was permitted to run.  Of course, it'd be nice to know (and eventually deliver that to the user as well) the error class the result is in..