WMS driver fails on skipping rows

[neteler]

WMS driver fails on skipping rows which would be needed to run the driver in external applications:

# gdal_wms.xml from http://www.gdal.org/frmt_wms.html
gdal_translate -outsize 20% 20% gdal_wms.xml lsat.tif
Input file size is 2666666, 1333333
0

It this point it "freezes" (strace shows the following):

...
select(8, [6 7], [], [], {0, 100000})   = 0 (Timeout)
poll([{fd=6, events=POLLIN|POLLPRI}], 1, 0) = 0
poll([{fd=7, events=POLLIN|POLLPRI}], 1, 0) = 0
select(8, [6 7], [], [], {0, 100000})   = 0 (Timeout)
...

Could this be fixed?

Thanks, Markus

[warmerdam]

Adam,

Thoughts?

[nowakpl]

Many actually,

1) no default timeout (or a very long one),  http://sydney.freeearthfoundation.com/gdalwms/r15671-2646.patch will use 300 seconds + configurable with <Timeout></Timeout>,

2) very large request, at 20% outsize expected output is ~400GB, GDAL: GDALDatasetCopyWholeRaster(): 533333*6 swaths, bInterleave=1, that translates into 652 tiles per RasterIO request, each request has to timeout before RasterIO returns error

3) old docs on  http://www.gdal.org/frmt_wms.html, new example config compatible with  http://wms.jpl.nasa.gov/tiled.html is on  http://sydney.freeearthfoundation.com/gdalwms/frmt_wms.html,

[rouault]

 http://sydney.freeearthfoundation.com/gdalwms/r15671-2646.patch commited in r15674

[rouault]

Docs updated in r15675 as for #2450

[nowakpl]

http://trac.osgeo.org/gdal/changeset/15674

added timeout with 300 second default