[GRASS5] migration from 5.3 to 5.7

[Radim Blazek]

Jaro Hofierka wrote:
> Hi,
> 
> I did some tests of reading point data in grass5.7 (500K points).
> I looks like the only problem is in reading 2D point data with 
> attributes (everytime I had to kill the process).
> However, reading 3D points in v.surf.rst is OK - it took around 10 secs
> to get them loaded. So finally we've got the solution at least partially.
> However, it is a bit complicated to produce 3D file from 2D site files
> because they must be converted (e.g. by awk to 3D format) and then
> can be imported as a 3D file using v.in.sites. The switch to force 3D 
> import would help a lot.
> Another interesting fact is that v.surf.idw is reading 2D point data 
> with attributes quite fast, however, it uses old site input functions. I 
> guess the vectors are read by some internal grass57 function that is 
> invoked by those site input functions.
> Maybe this function could be implemented in v.surf.rst as well. It would 
> solve our problem with 2D point data.
 >
> (dbf) v.surf.rst input=dmr100 elev=dmr100 (killed after 20 minutes)
> 2.86user 1.05system 20:02.44elapsed 0%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (705major+47732minor)pagefaults 0swaps

!!!
This was optimised in grass51/lib/rst/interp_float/vinput2d.c
in August (1.4) but the optimisation was simply reverted in 1.8 which 
was obviously based on CVS older then 1.4!!!

You cannot simply replace a file in CVS with your modified OLD version!!!

I have reverted vinput2d.c to 1.7.

Please check ALL the changes you have done! If your code was too old and 
if you have overwritten the previous changes, please revert and submit 
only your changes done in fresh CVS!!!

I am tired.

Radim

PS: On 10000 points this reduce the time from 1.344s to 0.093s