[Benchmarking] Styling gnis_pop

[Andrea Aime]

Hi,
the point layer we have on the benchmarking server is gnis_pop.

I've loaded it up on my local postgis to see what the styling potential
is.

Its structure, as reported by postgis, is:

  gid       | integer                | not null default
  fid       | bigint                 |
  name      | character varying(100) |
  class     | character varying(21)  |
  state     | character varying(2)   |
  county    | character varying(29)  |
  elevation | bigint                 |
  map       | character varying(30)  |
  the_geom  | geometry               |

The class is interesting, it contains quite a number of
different point types, 64 to be precise.

The distribution is as follows:
select class, count(*) as cnt from gnis_names_pg group by class order by cnt


          class         |  cnt
-----------------------+--------
  Sea                   |     11
  Isthmus               |     18
  Geyser                |    118
  Lava                  |    169
  Unknown               |    183
  Crater                |    239
  Plain                 |    283
  Slope                 |    361
  Arch                  |    445
  Arroyo                |    456
  Levee                 |    540
  Woods                 |    643
  Harbor                |    652
  Tunnel                |    697
  Bench                 |    723
  Rapids                |   1089
  Glacier               |   1090
  Reserve               |   1262
  Forest                |   1287
  Military (Historical) |   1772
  Pillar                |   2030
  Area                  |   2321
  Beach                 |   2322
  Range                 |   2395
  Falls                 |   2449
  Bend                  |   2783
  Channel               |   3911
  Basin                 |   4267
  Cliff                 |   4395
  Gut                   |   4542
  Oilfield              |   4858
  Bar                   |   5713
  Bridge                |   6213
  Swamp                 |   7386
  Gap                   |   8362
  Crossing              |  10088
  Flat                  |  10471
  Trail                 |  10537
  Hospital              |  11958
  Bay                   |  12732
  Ridge                 |  14946
  Cape                  |  16108
  Tower                 |  16656
  Island                |  18484
  Airport               |  19434
  Canal                 |  20570
  Post Office           |  25012
  Mine                  |  32915
  Spring                |  35847
  Well                  |  38232
  Civil                 |  38237
  Dam                   |  56929
  Park                  |  62308
  Building              |  65132
  Lake                  |  68012
  Summit                |  70223
  Valley                |  71857
  Reservoir             |  75260
  Cemetery              | 126167
  Locale                | 132971
  Populated Place       | 180279
  School                | 181041
  Church                | 188695
  Stream                | 241440

I guess we could take some 10 of the most popular
classes and have them symbolized with a png or
ttf marker and have a label drawn below it.
How does this sound? With some varied zoom
levels it should exercise also label conflict
detection.

The inherent filtering would make it evident
what system can do to filter data (I hear that
ESRI can index the DBF attributes for example).
Database wise we would add an index on the
"class" column.

Oh, the data set is complete USA wise and contains
almost 2 million points, to make it comparable with
the roads layers maybe we should filter out only
the points in Texas? That would leave us with a much
smaller data set, only 95k points and the following
distribution:

          class         |  cnt
-----------------------+-------
  Crater                |     1
  Bench                 |     1
  Slope                 |     2
  Tunnel                |     2
  Arch                  |     5
  Rapids                |     5
  Plain                 |     6
  Forest                |     9
  Reserve               |    11
  Woods                 |    17
  Arroyo                |    19
  Harbor                |    19
  Pillar                |    21
  Beach                 |    27
  Area                  |    27
  Falls                 |    29
  Bar                   |    55
  Mine                  |    56
  Post Office           |    57
  Crossing              |    61
  Range                 |    63
  Basin                 |    65
  Military (Historical) |    67
  Levee                 |    87
  Ridge                 |   110
  Bridge                |   130
  Channel               |   146
  Gap                   |   157
  Flat                  |   180
  Cliff                 |   185
  Swamp                 |   231
  Gut                   |   259
  Bend                  |   259
  Island                |   261
  Civil                 |   262
  Cape                  |   279
  Bay                   |   283
  Canal                 |   299
  Trail                 |   504
  Hospital              |   579
  Well                  |   942
  Tower                 |  1052
  Spring                |  1243
  Oilfield              |  1294
  Airport               |  1757
  Lake                  |  1780
  Summit                |  2117
  Valley                |  2844
  Building              |  3795
  Park                  |  4008
  Dam                   |  5947
  Cemetery              |  6016
  Locale                |  7980
  Populated Place       |  8511
  Reservoir             |  8542
  School                |  8756
  Stream                | 11640
  Church                | 12072

Cheers
Andrea