[GRASS-dev] vector large file support
Markus Metz
markus.metz.giswork at googlemail.com
Wed Feb 11 06:54:28 EST 2009
Glynn Clements wrote:
> Markus Metz wrote:
>
>
>> BTW, display of grass7 with cairo appears to be half the speed of
>> grass65, I think it was faster last week (can't remember the revision).
>> Not sure if this justifies a complaint.
>>
>
> Almost everything related to display has changed between 6.x and 7.0.
>
> One relatively recent change that may cause a slow-down is that the
> display library no longer tries to automatically simplify vectors
> (reduce consecutive short line segments with a single segment). This
> could have a significant impact when viewing detailed vectors at low
> zoom.
>
OK. It does have a significant impact when viewing detailed vectors at
low zoom (relatively detailed: 25MB coor, 31MB topo, not that large but
complicated topology). I can't see a speed difference with small
vectors. That just got me by surprise because I was used to a 7.0
display that's faster, better, nicer than the 6.x display. Usually, the
screen resolution can't represent that detail (same for raster map
display), but I guess it is more correct to try to display all detail.
BTW, I'm testing LFS enabled vector libs in 7.0. So far it compiles as
good or bad as before on Linux 32bit with and without --enable-largefile
and on Linux 64bit with --enable-largefile. The test in
lib/vector/diglib is not passed when compiling on Linux64bit without
--enable-largefile, but the libs work anyway. I did not have to increase
any of the version numbers, the vectors are fully backwards compatible,
tested with 6.4.0RC3. It would be safer to change that though, and
change the layout of the topo and cidx file.
IMHO LFS support for vector libs is not necessarily good, currently. In
order to manipulate large vectors, a very fast CPU, fast hard drives and
lots of RAM is needed, or a lot of patience, because it can take days or
weeks to run modules with large vectors as input. That speed issue may
change in the future, and I would regard LFS support in the vector libs
as an investment in the future. Maybe I should rather try to fix bugs
than to add new features...
More information about the grass-dev
mailing list