this bug's URL: http://intevation.de/rt/webrt?serial_num=1721
-------------------------------------------------------------------------
Subject: s.in.ascii should deal with two column input better
Platform: GNU/Linux/i386
grass obtained from: Mirror of Trento site
grass binary for platform: Compiled from Sources
GRASS Version: 5.0.0
s.in.ascii doesn't deal with two column data very well and fails with a cryptic error message.
example:
test.prn:
2101781 5619950
2101781 5619900
2101781 5619850
2101781 5619800
2101781 5619750
2101781 5619700
2101781 5619650
GRASS:~ > s.in.ascii sites=test input=test.prn
WARNING: error scanning floating point attribute
WARNING: error scanning floating point attribute
WARNING: error scanning floating point attribute
WARNING: error scanning floating point attribute
WARNING: error scanning floating point attribute
WARNING: error scanning floating point attribute
WARNING: error scanning floating point attribute
What's a new user to think! This (along with v.in.shape, et al) is probably the first things they try. Spend a few hours ripping their hair out trying different file formats and then give up on GRASS all together as "too hard, too crude, doesn't work".. and go away feeling bad.
From the interactive input:
Enter sites, one per line, in the format:
east north attributes
When finished, type: end
location attributes> 2 3
WARNING: error scanning floating point attribute
location attributes> 2 3 4
location attributes> 1 2
location attributes> 1 2 3
location attributes> 12 13 13 13
location attributes> 1 1
location attributes> 1
** invalid format **
location attributes> end
which is interesting because it accepts e,n input and auto-increments cat #s, but only after a first attribute is defined.
Rather than just making a more helpful error message, what I think should really happen is if only a easting and northing are detected on the first line, s.in.ascii should accept it and assign and increment category numbers automatically. This already happens so I can't imagine it would be too hard to scan the first line of input ala argc==2 of whatever.
thanks,
Hamish (who'll leave messing with the important modules to the more experienced)
[and who just updated the man page for this module, which was just compounding the problem with wrong information]
-------------------------------------------- Managed by Request Tracker