[GRASSLIST:6046] Questions

GRASS GRASS user
1

Well, i get the prize for the most descriptive subject...
Anyway, i use GRASS for some time now. I used the binaries and with your help through this list i have only minor problems.
But I decided to compile the 5.0.1 source yesterday.
Compilation was ok (i directed with options most things to /sw/include and /sw/lib).
The only warning in the configure script was fot the iconv() and gettext() function. (undefined functions).
Now i encounter a problem.I cant start tcltk gui.
When i use tcltkgrass & from within grass i get the message
GRASS 5.0.1 > /usr/local/grass5/bin/tcltkgrass: exec: wish: not found

DO i have to edit tcttk sctipt in some way?

Thank you
Konstantinos Theofilis

2

Konstantinos,

there is an entry about this in the FAQ. See
http://www.geog.uni-hannover.de/grass/faq/wishnotfound.html

Regards,
Vincent Schut.

On Monday 14 April 2003 15:23, Edmund Blackadder wrote:
...

When i use tcltkgrass & from within grass i get the message
GRASS 5.0.1 > /usr/local/grass5/bin/tcltkgrass: exec: wish: not found

DO i have to edit tcttk sctipt in some way?

Thank you
Konstantinos Theofilis

--
______________________________________
Vincent Schut (schut@sarvision.com)
Sarvision B.V.
Wageningen, The Netherlands
www.sarvision.com

3

I have compiled nviz before by fudging the tkint header file required by togl.c when I had a major/minor version for TclTknot listed among the included tkIntX.X.h files

The newest grass source code (5.0.1) doesn't seem to have an option for tk 8.4.2 and won't let me "cob" the solution...or I have forgotten what happened last time I did this.

I have TclTk 8.4.2 installed as a package under slackware in /usr/lib (and /usr/include etc) as the default tcl setup

I have TclTk8.0.5 compiled from source and installed in /usr/local/tcltk8 for use with TkDesk...it is included in my ld.so.conf

I get compiler fatal errors for nviz regarding missing tkint8.4.2 OR with "duplicate declarations" for various methods depending upon how I attempt to handle it.

What am I missing?

4

Thanks,
My wish command was in /sw/bin because of Fink
S i did
setenv GRASS_WISH /sw/bin/wish

Konstantinos Theofilis

On Monday, April 14, 2003, at 02:40 PM, Vincent Schut wrote:

Konstantinos,

there is an entry about this in the FAQ. See
http://www.geog.uni-hannover.de/grass/faq/wishnotfound.html

Regards,
Vincent Schut.

On Monday 14 April 2003 15:23, Edmund Blackadder wrote:
...

When i use tcltkgrass & from within grass i get the message
GRASS 5.0.1 > /usr/local/grass5/bin/tcltkgrass: exec: wish: not found

DO i have to edit tcttk sctipt in some way?

Thank you
Konstantinos Theofilis

--
______________________________________
Vincent Schut (schut@sarvision.com)
Sarvision B.V.
Wageningen, The Netherlands
www.sarvision.com

5

Edmund Blackadder wrote:

Well, i get the prize for the most descriptive subject...
Anyway, i use GRASS for some time now. I used the binaries and with
your help through this list i have only minor problems.
But I decided to compile the 5.0.1 source yesterday.
Compilation was ok (i directed with options most things to /sw/include
and /sw/lib).
The only warning in the configure script was fot the iconv() and
gettext() function. (undefined functions).

These are harmless.

iconv() is only used by d.text.freetype; if you don't have iconv(),
d.text.freetype still works (assuming it was actually built, which
requires FreeType support), but it can only handle Western
(ISO-8859-1, aka ISO-Latin1) text.

gettext() is only used if you use --with-nls to get natural language
support. Currently, only Russian is supported, and only for parts of
GRASS (libgis, libraster, the PostgreSQL modules, and a couple of
other programs).

Neither of these are related to the tcltkgrass problem. Either set the
environment variable GRASS_WISH to the name of your "wish" program (it
may include the version number, e.g. "wish8.3"), or create a symlink
called "wish" pointing to the actual wish program.

--
Glynn Clements <glynn.clements@virgin.net>

6

System: Mac OSX 10.2.4
Grass Version: OpenOSX binary 5.0 with OpenOSX update 0.1
Problem:
When I attempt to use i.pca under either xterm or the tcltk interface Grass parses the function as
r.in.gdal, and exits with an error.

for example the function call:

i.pca -help

Yields the following text which clearly refers to r.in.gdal::
<<snip>>
Description:
  Import GDAL supported raster file into a binary raster map layer.

Usage:
  i.pca [-oe] input=name [band=value] output=name [target=name]
    [title="phrase"] [location=name]

Flags:
   -o Override projection (use locations projection)
    <<snip>>

Is this a compile problem with the OpenOSX binary, or local to my install, or is a universal bug ?

How do i fix i.pca ?

-john chesnut

7

Apparently it was a problem with using TclTk 8.4.2...I recompiled after uninstalling the slackware TclTk842 package and fixing my symlinks and lib paths and nviz compiled without error.

Andrew wrote:

I have compiled nviz before by fudging the tkint header file required by togl.c when I had a major/minor version for TclTknot listed among the included tkIntX.X.h files

The newest grass source code (5.0.1) doesn't seem to have an option for tk 8.4.2 and won't let me "cob" the solution...or I have forgotten what happened last time I did this.

I have TclTk 8.4.2 installed as a package under slackware in /usr/lib (and /usr/include etc) as the default tcl setup

I have TclTk8.0.5 compiled from source and installed in /usr/local/tcltk8 for use with TkDesk...it is included in my ld.so.conf

I get compiler fatal errors for nviz regarding missing tkint8.4.2 OR with "duplicate declarations" for various methods depending upon how I attempt to handle it.

What am I missing?

8

John Chesnut wrote:

System: Mac OSX 10.2.4
Grass Version: OpenOSX binary 5.0 with OpenOSX update 0.1
Problem:
When I attempt to use i.pca under either xterm or the tcltk
interface Grass parses the function as
r.in.gdal, and exits with an error.

for example the function call:

i.pca -help

Yields the following text which clearly refers to r.in.gdal::
<<snip>>
Description:
  Import GDAL supported raster file into a binary raster map layer.

Usage:
  i.pca [-oe] input=name [band=value] output=name [target=name]
    [title="phrase"] [location=name]

Flags:
   -o Override projection (use locations projection)
    <<snip>>

Is this a compile problem with the OpenOSX binary, or local to my
install, or is a universal bug ?

This suggests that the links in the bin directory (e.g.
/usr/local/grass5/bin) are pointing at a binary which is effectively
r.in.gdal.

All of the "files" in the bin directory (except for r.agnps50.run,
tcltkgrass and nviz) should be hard links to $GISBASE/etc/front.end:

  $ ls -i /usr/local/grass5/bin
  8224858 bmif_to_cell
  8224858 d.3d
  8224858 d.area
  8224858 d.ask
  8224858 d.barscale
  8224858 d.colorlist
  8224858 d.colormode
  ...
  $ ls -i /usr/local/grass5/etc/front.end
  8224858 /usr/local/grass5/etc/front.end

The front.end program invokes the appropriate program from either
etc/bin/cmd or etc/bin/inter, depending upon whether any command-line
arguments were used and whether a separate interactive version exists.

The most likely cause for this situation is that something attempted
to replace bin/r.in.gdal (rather than etc/bin/cmd/r.in.gdal) with an
r.in.gdal binary, and did so by overwriting the existing front.end
program, so front.end is now effectively r.in.gdal.

The easiest way to test this theory is to start GRASS then try to run
etc/front.end directly, e.g.

  GRASS:~ > /usr/local/grass5/etc/front.end help

Normally, you should get an error message:

  ERROR: program 'front.end' cannot be executed because: Neither a command
         line (cmd) or interactive (inter) version was found

If you get the r.in.gdal help message instead, then my theory is
correct. In which case, try re-installing from the CD. If that doesn't
work, there's something wrong with the CD.

--
Glynn Clements <glynn.clements@virgin.net>