Hi all,
about two weeks ago Curt started preparations for a scenery regeneration
run using the new shapefile data from Martin Spott's Terrain Data
Database. In order to be able to accept custom scenery modifications to
the standard scenery we need to get this part working.
This is what he reported to me:
I haven't had a chance to dig into this yet (and I'm not sure I will
today) but when processing the rivers_stream database, it reports
760648 records. However, the process quit peacefully after
processing only about 72000 records (i.e. < 10% of the reported
number of records). Any ideas. I didn't see any sort of
segfault/coredump type message.
I was unable to reproduce the problem up to now and we're kinda stuck.
I'm therefore searching for helpers who can get TerraGear running on
their system and try to reproduce the bug. You can use
src/Prep/ShapeFile/process.sh for this.
Download the shapefiles
ftp://ftp.ihg.uni-duisburg.de/FlightGear/TGShapes/rivers_stream.tar.bz2
and optionally
ftp://ftp.ihg.uni-duisburg.de/FlightGear/TGShapes/landmass_default.tar.bz2
and untar them into a new directory. Change process.sh to have SHAPEBASE
point at these shapefiles and WORKBASE at a place where you want to put
the results.
Try it only with rivers_stream.tar.bz2 for starters, but I'm not sure
whether the previous instance of shapedecode (decoding landmass) already
mixes up the working directory, which could cause the crashes as well.
WARNING: This test run will take many GBs of your free disk space. The
last time I tried, I wasn't yet at the 72000 records mark and my disk
was full (previously 5GB free).
So far we found out that the polygons for lines touching the date line
(180 deg longitude E/W) get mixed up, as the widened polygons for these
lines cross the date line. However, we were not able to find out whether
this has anything to do with the crash/termination Curt described.
So far Curt saw it crash on these records: 72447, 194774, 196576
Thanks in advance.
Cheers,
Ralf
_______________________________________________
Flightgear-devel mailing list
[email protected]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d
