SDTS Import name proposal:
The following scheme allows transfer between DOS and UNIX, and
accomodates
multiple manifold (layer) transfers.
Example:
SDTS dataset has filenames of the form PRNMxxxx.DDF
There are 7 different layers or manifolds, all for the
same partition.
Assume thee data is being imported using the root name "junk"
1. grass vector map names:
11 character limit plus 1-3 character extension. use the extension to
distinguish
layers:
e.g.
junkette_1, junkette_2, junkette_23
(or wmhy_del.434)
2. database names (these will be "rdbms-ready" ascii files for SDTS
attributes
and the GRASS-RDBMS "link" files):
a. for regular attribute tables:
4-character SDTS prefix +
4-char module name + .db extension:
e.g. PRNMAP01.db
HY01B011.db
b. GRASS/database link tables:
(1) single maps
grass layer_name + .db
(1) multiple maps first letter of grass layer-name
[ + number of layer ]
+ sdts-prefix + .db
e.g.
if layer name is "junkette" and there are 7 surfaces:
j_1PRNM.db
j_2PRNM.db
j_3PRNM.db
j_4PRNM.db
j_5PRNM.db
j_6PRNM.db
j_7PRNM.db
c. composite link tables
PRNMffel.db
PRNMffat.db
PRNMobat.db
Special cases:
1. Simple GRASS-type import, i.e., where there is only a single composite
link-attribute database file. (this will apply to all datasets with
1-to-1 obj-attr relations and a single attribute file)
use user-specified output name for both grass map and attribute
file, as follows:
grass map:
rdbms-ready file:
junkette
junkette.db
2. If either the SDTS prefix, or the output maplayer name chosen by the
user lacks a leading alpha character (e.g., as in TIGER/SDTS's 2500
prefix),
the letter 'T' is arbitrarily prepended to the db-ready filenames in the
db create and load scripts. Informix and other SQL rdbms's need tables
and databases to begin with a letter.
so, the db table for "2500AP57.db" will be "T2500AP57".