xyzm (rebased)

[mbredif]

Rebased and cleaned up version of the xyzm PR. This PR is to gather discussion, not for immediate merging.

[elemoine]

PR base changed from dev to master.

[pblottiere]

@mbredif

Good work! I am looking forward to have this PR mergde in master!

I wanted to try your work in order to remove the pc_filterbetween on Z dimension in some queries. But I came across some issues. I know that you're still working on it, so it's maybe 'normal' for now.

Firstly when I'm using autotools, I have this error during the install step:

/usr/bin/install -c -m 644 .//pointcloud_postgis.control '/usr/local/share/postgresql/extension/'
/usr/bin/install -c -m 644 .//pointcloud_postgis--1.1.0.sql  '/usr/local/share/postgresql/extension/'
/usr/bin/install: cannot stat './/pointcloud_postgis--1.1.0.sql': No such file or directory
/usr/local/lib/postgresql/pgxs/src/makefiles/pgxs.mk:114: recipe for target 'install' failed
make[1]: *** [install] Error 1
make[1]: Leaving directory '/home/blottiere/devel/packages/3dexplorer/pointcloud_mbredif_xyzm_rebased/pgsql_postgis'
Makefile:3: recipe for target 'install' failed
make: *** [install] Error 2

With Cmake, the install step seems to be working fine. Then, I have the next error when I want to create an index using gist on geometry(pa):

psycopg2.ProgrammingError: function geometry(pcpatch) does not exist
LINE 1: ... if not exists patchs_pa_idx on patchs using gist(geometry(p...
                                                             ^
HINT:  No function matches the given name and argument types. You might need to add explicit type casts.

Thank you!

[mbredif]

Firstly when I'm using autotools, I have this error during the install step

I confess I use cmake, so autotools may be lagging behind. The culprit should be in mbredif/pointcloud@15da35e. PR for autool support is welcome :)

Then, I have the next error when I want to create an index using gist on geometry(pa):

For consitency, I proposed to change the SQL API in mbredif/pointcloud@62acf84. The geometry cast is still in place, but I changed geometry(pa) to a more explicit PC_Envelope(pa) which is now a valid postgis geometry rather than a bytea.

[pblottiere]

I confess I use cmake, so autotools may be lagging behind. The culprit should be in mbredif/pointcloud@15da35e. PR for autool support is welcome :)

It's done.

For consitency, I proposed to change the SQL API in mbredif/pointcloud@62acf84. The geometry cast is still in place, but I changed geometry(pa) to a more explicit PC_Envelope(pa) which is now a valid postgis geometry rather than a bytea.

The index creation seems OK with PC_Envelope function. Thank you.

[mbredif]

Thanks for the PR @pblottiere.
I am eager to hear about your performance results ;)