You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the last meeting we discussed how the styles are parsed.
Currently if two styles are completely the same they share the style index.
With the introduction of stops we want to blend two styles between two zoom levels. To do this more efficiently we want that if we blend two styles together the next style hast to be one index above the current style.
This creates the problem that it is not so simple anymore to determine if two styles are the same. (since we now have to parse all the steps first before we can decide whether to create new styles or reuse old ones)
We therefore currently always create new styles (even if the style already existed somewhere before).
test how many styles could potentially be combined (in order to determine the importance of this)
if necessary implement the merging
The text was updated successfully, but these errors were encountered:
commit dbdd34187bed5ba6e9e35c7d6befafbcac7fda45 reuses styles if they are reused in subsequent zoom levels.
Meaning style for zoom 12-18, 15-18 has stops -> 12-15 uses the same style and does not need to be declared multiple times. The Style for zoom level 15 has blending enabled, but this is declared in the style index -> it still can use the same style that 12-14 is using.
here are the current number of styles that are stored in the style buffer:
style
reusing any
all separate
reusing subsequent
openmaptiles
821
2401
1230
qwant
462
1201
735
bright
497
1255
821
basemap
205
1650
455
While there are still many styles that are currently duplicated, this improvement nevertheless already reduced duplicate styles by a lot. Furthermore by capturing and encoding more values from the style.json (e.g. dashes) the number of "reusing any" should further increase to a value that is closer to the "reusing subsequent" variation.
In the last meeting we discussed how the styles are parsed.
Currently if two styles are completely the same they share the style index.
With the introduction of stops we want to blend two styles between two zoom levels. To do this more efficiently we want that if we blend two styles together the next style hast to be one index above the current style.
This creates the problem that it is not so simple anymore to determine if two styles are the same. (since we now have to parse all the steps first before we can decide whether to create new styles or reuse old ones)
We therefore currently always create new styles (even if the style already existed somewhere before).
The text was updated successfully, but these errors were encountered: