[Marble-devel] Review Request 123239: Optimize linestring rendering via automatically assigned detail level values.

[Torsten Rahn]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/123239/#review78497
-----------------------------------------------------------


Issues with this patch:
- Antarctica isn't rendered correctly
- It might be useful to change the algorithm so that it properly handles cases more decently with 3-4 nodes

- Torsten Rahn


On April 3, 2015, 1:34 nachm., Torsten Rahn wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/123239/
> -----------------------------------------------------------
> 
> (Updated April 3, 2015, 1:34 nachm.)
> 
> 
> Review request for Marble, Bernhard Beschow, Dennis Nienhüser, Marius Stanciu, and Thibaut Gridel.
> 
> 
> Repository: marble
> 
> 
> Description
> -------
> 
> Allow for automatically assigning detail values to the nodes inside a linestring.
> This mechanism is used for PN2 parsing in this patch.
> Taking the detail level into account this approach improves geometry layer performance by 20% on my i7 based machine.
> This patch also fixes the "backwards" semantical meaning of ViewportParams::resolve()
> In the next step this optimization could also be applied during KML and OSM parsing.
> 
> 
> Diffs
> -----
> 
>   src/lib/marble/ViewportParams.cpp bca8ea9 
>   src/lib/marble/geodata/data/GeoDataCoordinates.h 610b423 
>   src/lib/marble/geodata/data/GeoDataLineString.h 4e4e3da 
>   src/lib/marble/geodata/data/GeoDataLineString.cpp 8864518 
>   src/lib/marble/geodata/data/GeoDataLineString_p.h 2c7b9ec 
>   src/lib/marble/projections/AbstractProjection.cpp 591adc0 
>   src/lib/marble/projections/AbstractProjection_p.h 7bbfcb5 
>   src/lib/marble/projections/AzimuthalProjection.cpp d6e7376 
>   src/lib/marble/projections/CylindricalProjection.cpp 42345ab 
>   src/plugins/runner/pn2/Pn2Runner.cpp fd363c8 
> 
> Diff: https://git.reviewboard.kde.org/r/123239/diff/
> 
> 
> Testing
> -------
> 
> Tested manually using the Atlas map, Plain map and Political map without apparent regressions. "make test" runs fine.
> Checked the geometry layer values by starting Marble with --runtimeTrace. On average the msec values are about 20% lower on my machine (and I hope that improvement is even better on slower CPUs).
> Ideas for unit tests regarding this patch are welcome. ;-)
> 
> 
> Thanks,
> 
> Torsten Rahn
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/marble-devel/attachments/20150404/e6de7a50/attachment-0001.html>