Ticket #7 (closed defect: fixed)

Opened 11 years ago

Last modified 11 years ago

Reprojection problem with whole world extents

Reported by: dmorissette Owned by: warmerdam
Priority: high Milestone:
Component: GDAL Support Version: 3.5
Severity: normal Keywords:
Cc:

Description 

Requesting a map with BBOX=-180,-90,180,90 produces an error in the GDAL 
resampling code about lat/lon values exceeding the allowed limits.

Similar thing happens with vectors, but instead of getting an error we get 
points reprojected at the other end of the world.

Here is Frank's reply:

This is a result of the PROJ.4 pj_transform() call failing.  I pass the
error on up with msSetError(). 

Clearly this needs to be fixed, and I am sure something can be done.  

Is there a bug report in bugzilla on these two issues?  Who will make them
up?

As for the vector side of things, I don't know the data paths well.  This
must not be anything new though right?  I notice that the map bounds
are 154 to -154 in latitude when you request a world view.

Change History

Changed 11 years ago by fwarmerdam

  • status changed from new to assigned
  • rep_platform changed from PC to All
  • component changed from WMS Compatibility to GDAL Support
  • op_sys changed from Windows 98 to All 
I am reassigning this to the GDAL support component, and am planning to 
tackle only the GDAL aspect for now.   It might be good to create a separate
bug report for the vector issues, since they are only loosely related to the
raster ones.

Changed 11 years ago by fwarmerdam 

The following demonstrates the basic problem (in traditional MapServer URL
format) more briefly. 

http://gdal3.dreadfest.com/cgi-bin/mapserv?mode=browse&zoomdir=1&zoomsize=2&imgxy=299.5+299.5&imgext=-180+90+180+-90&map=%2Fusr%2Flocal%2Fapache%2Fhtdocs%2Frot_ortho%2Fdemo.map&savequery=true&mapext=shapes&img.x=106&img.y=184

Changed 11 years ago by fwarmerdam

  • status changed from assigned to closed
  • resolution set to fixed 
I have made modifications to mapresample.c that seem to resolve the 
problem.  In particular, if the map extents fail to reproject into the
projection of the source image when computing the portion of the source
image to load we just load the entire source image (not necessarily at
full resolution). 

This whole area is likely to become a more severe problem when I start
using approximated transformations for the reprojection warp as these
approximations will have to be built very carefully based only on
valid regions.  

Note: I have only fixed raster problems ... there are still vector "issues".

Changed 11 years ago by dmorissette 

I have filed bug #15 for the vector issues from this bug report:
http://mapserver.gis.umn.edu/bugs/show_bug.cgi?id=15
Note: See TracTickets for help on using tickets.