[GRASS5] compiling beta10 on IRIX and CYGWIN/WIN32

GRASS GRASS developer
1

Hi 2 all,

i want to report some problems i found with GRASS5.0beta10 (not the CVS
version!) on IRIX 6.5 and CYGWIN.

I am tired of fixing the same problems again and again. So i want to ask
everyone to remember:
The Gmakefile lib order _must_ always be:
$(LIBS) $(XDRLIB) $(MATHLIB)
Otherwise on IRIX it won't compile. If you are linking to the gis lib,
you must always include $(MATHLIB), as gis references floor, ceil, etc.
from the math lib. If you are not sure, add $(XDRLIB) $(MAHTLIB), it
won't hurt.

I had to heavily edit the header file for CYGWIN, there are still a lot
of problems with include and lib search paths etc.
And the command line/text mode interface is not working any more.

Please see the attached file for the problems in detail.

I have currently no time to work on the fixes myself. I will concentrate
on the XDRIVER/fifo/ipc stuff.

Andreas
--
Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net

(attachments)

bugs-grass5.beta10.txt (2.23 KB)

2

On Sat, Dec 09, 2000 at 01:07:58PM +0100, Andreas Lange wrote:

Hi 2 all,

i want to report some problems i found with GRASS5.0beta10 (not the CVS
version!) on IRIX 6.5 and CYGWIN.

I am tired of fixing the same problems again and again. So i want to ask
everyone to remember:
The Gmakefile lib order _must_ always be:
$(LIBS) $(XDRLIB) $(MATHLIB)
Otherwise on IRIX it won't compile. If you are linking to the gis lib,
you must always include $(MATHLIB), as gis references floor, ceil, etc.
from the math lib. If you are not sure, add $(XDRLIB) $(MAHTLIB), it
won't hurt.

Hi Andreas,

actually I don't see the problem:

Is the order:
$(LIBS) $(MATHLIB) $(XDRLIB)

causing problems? If yes, we have to change 60% (or more) of the Gmakefiles.
If so, please modify SUBMITTING. I have a script here which could do
the change of order.

Generally I was sure that 99% of the Gmakefiles are ok. Obviously I am
wrong?

How does Justin manage if not (he is using Irix, too)?

As Linux doesn't complain if $(XDRLIB) is missing, what is the rule for this
requirement? Then I can check all Gmakefiles...

Little confused (as beta10 is rather identical to CVS at time),

Markus

PS: I'll look into problems mentioned below:

I had to heavily edit the header file for CYGWIN, there are still a lot
of problems with include and lib search paths etc.
And the command line/text mode interface is not working any more.

Please see the attached file for the problems in detail.

I have currently no time to work on the fixes myself. I will concentrate
on the XDRIVER/fifo/ipc stuff.

Andreas
--
Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net

GRASS5.0beta10:

IRIX:
g.version has double \n's (fix is on the way)

gis_set.tcl: -state disabled instead of -state disable
gis_set.tcl: string compare -nocase: not supported
although this is supported from tk8.2 on, it is not
ok to use this if we claim tk8.0 compatibility.

when converting the man pages:
many error messages with awf formatter:
. DE not available (or such)

g.manual from tcltkgrass:
Illegal option -- P
Usage: man ...
g.manual entries=d.area -f:
Usage:
g.manual .....

this is all messed up, rman: please do not ask the user to install
something from within Usage info. This will confuse "ordinary" users.

CELL driver seems to work on IRIX in latlon. (colors sometimes incorrect)
in spearfish/UTM the intermittend white lines still appear.

r.mapcalc:
couldn't reproduce the behaviour Anantha Prasad reported.
This BUG is gone.
The problem with the reduce/reduce conflicts on compiling
can IMHO be ignored, as the output of the parser is parsed again
with a C program. To my knowledge the parser is ok and
i know of no errors.
This is the same with r3.mapcalc.

r.in.png does not work
(r.out.png, then r.in.png with this image gives me a totally white
raster on IRIX). r.stats says: 100% * (NULL)

r.in.tiff on IRIX:
r.in.tiff [1591] - out of logical swap space during exec - see swap(1M)
killed.
argh, i have 128 MB RAM and another 128 M RAM swap.
Must be some sort of segfault. This happens when invoking only
r.in.tiff, without any arguments!

wrong order of libs: $(MATHLIB) in
src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile

src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile

CYGWIN:
grass5
/usr/local/grass5/etc/Init.sh: not found
does not start with and without any commandline flag (-text, -tcltk etc.)

COMPLETELY BROKEN NOW!

GENERAL:

g.help:
PAGEUP/PAGEDOWN: doesn't work on every terminal.
E. g. on IRIX it is broken.
This needs to take the termcap into account. It is not possible
to hardcode the keycodes into the program.

t.roads.inf from spearfish database is broken/wrong format.
no cat support (informix db?)

tests that i performed:
for f in `g.mlist type=rast` ; do
  d.erase white
  echo "plotting $f"
  d.rast $f
  sleep 2
done
and with d.vect, d.area, d.site etc.

--
andreas lange, 12/2000

--
Dipl.-Geogr. Markus Neteler * University of Hannover
Institute of Physical Geography and Landscape Ecology
Schneiderberg 50 * D-30167 Hannover * Germany
Tel: ++49-(0)511-762-4494 Fax: -3984

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

3

Hi,
  i had the following error message when compiling
the beta 10:

[root@snarf vask]# gmake5
  SRC = /usr/local/src/grass5.0beta10/src
  CMD = /usr/local/src/grass5.0beta10/src/CMD
  HEADER = head.i686-pc-linux-gnu
  ARCH = i686-pc-linux-gnu
  GISBASE = /usr/local/grass5
  VERSION = 5.0beta10 December 2000
#################################################################
/usr/local/src/grass5.0beta10/src/libes/vask
  make -f OBJ.i686-pc-linux-gnu/make.rules gcc -g -O2 -DMAKE_VASK_LIB
-I/usr/include/ncurses.h -I/usr/local/src/grass5.0be ta10/src/include -c
V_call.c -o OBJ.i686-pc-linux-gnu/V_call.o
/usr/include/ncurses.h: Not a directory

which comes from the line:

EXTRA_CFLAGS = -DMAKE_VASK_LIB -I/usr/include/ncurses.h

in src/libes/vask/Gmakefile

I removed the "/ncurses.h" part and all works correctly now.

          Federico

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

4

Hi Markus, hi others,

sorry for the confusing message.
The problem is with the two Gmakefiles mentioned later in the text.
Only these two Gmakefiles cause problems because of missing $(MATHLIB)
_or_ $(MATHLIB) _before_ $(GISLIB), which makes the IRIX linker fail.
So the order should be:
$(LIBES) $(WHATEVER) $(GISLIB) $(MATHLIB) $(XDRLIB)

I just was a little bit frustrated because both programs compiled with
beta8. So everyone should remember not to delete/reorder the LIBES. Then
the Gmakefiles should be 100% correct.

The $(XDRLIB) may be useless on linux, but is needed for win32.

I am too overworked right now to look into the Cygwin problem with the
Init.sh. This should be fixed, as the windows port will not work on
commandline :-(. No graphics is bad, but setting the GRASS environment
by hand is nearly impossible.

cu,

Andreas

Markus Neteler wrote:

Hi Andreas,

actually I don't see the problem:

Is the order:
$(LIBS) $(MATHLIB) $(XDRLIB)

causing problems? If yes, we have to change 60% (or more) of the Gmakefiles.
If so, please modify SUBMITTING. I have a script here which could do
the change of order.

Generally I was sure that 99% of the Gmakefiles are ok. Obviously I am
wrong?

How does Justin manage if not (he is using Irix, too)?

As Linux doesn't complain if $(XDRLIB) is missing, what is the rule for this
requirement? Then I can check all Gmakefiles...

Little confused (as beta10 is rather identical to CVS at time),

Markus

PS: I'll look into problems mentioned below:

> wrong order of libs: $(MATHLIB) in
> src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile
>
> src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile

--
Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

5

No Markus, you do not have to change 60% of the Gmakefiles.
if beta10 comes from the cvs.

I modified all the files to be $(LIBS) $(XDRLIB) $(MATHLIB)
when I first started. It is usually new files that potentially may
have incorrect order. The rules depends on which LIBS are
included. Some require XDRLIB and some don't.

John Huddleston

----- Original Message -----
From: "Markus Neteler" <neteler@geog.uni-hannover.de>
To: <grass5@geog.uni-hannover.de>
Sent: Saturday, December 09, 2000 11:06 AM
Subject: Re: [GRASS5] compiling beta10 on IRIX and CYGWIN/WIN32

On Sat, Dec 09, 2000 at 01:07:58PM +0100, Andreas Lange wrote:
> Hi 2 all,
>
> i want to report some problems i found with GRASS5.0beta10 (not the CVS
> version!) on IRIX 6.5 and CYGWIN.
>
> I am tired of fixing the same problems again and again. So i want to ask
> everyone to remember:
> The Gmakefile lib order _must_ always be:
> $(LIBS) $(XDRLIB) $(MATHLIB)
> Otherwise on IRIX it won't compile. If you are linking to the gis lib,
> you must always include $(MATHLIB), as gis references floor, ceil, etc.
> from the math lib. If you are not sure, add $(XDRLIB) $(MAHTLIB), it
> won't hurt.
Hi Andreas,

actually I don't see the problem:

Is the order:
$(LIBS) $(MATHLIB) $(XDRLIB)

causing problems? If yes, we have to change 60% (or more) of the Gmakefiles.
If so, please modify SUBMITTING. I have a script here which could do
the change of order.

Generally I was sure that 99% of the Gmakefiles are ok. Obviously I am
wrong?

How does Justin manage if not (he is using Irix, too)?

As Linux doesn't complain if $(XDRLIB) is missing, what is the rule for this
requirement? Then I can check all Gmakefiles...

Little confused (as beta10 is rather identical to CVS at time),

Markus

PS: I'll look into problems mentioned below:

> I had to heavily edit the header file for CYGWIN, there are still a lot
> of problems with include and lib search paths etc.
> And the command line/text mode interface is not working any more.
>
> Please see the attached file for the problems in detail.
>
> I have currently no time to work on the fixes myself. I will concentrate
> on the XDRIVER/fifo/ipc stuff.
>
> Andreas
> --
> Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
> Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net

> GRASS5.0beta10:
>
> IRIX:
> g.version has double \n's (fix is on the way)
>
> gis_set.tcl: -state disabled instead of -state disable
> gis_set.tcl: string compare -nocase: not supported
> although this is supported from tk8.2 on, it is not
> ok to use this if we claim tk8.0 compatibility.
>
> when converting the man pages:
> many error messages with awf formatter:
> . DE not available (or such)
>
> g.manual from tcltkgrass:
> Illegal option -- P
> Usage: man ...
> g.manual entries=d.area -f:
> Usage:
> g.manual .....
>
> this is all messed up, rman: please do not ask the user to install
> something from within Usage info. This will confuse "ordinary" users.
>
> CELL driver seems to work on IRIX in latlon. (colors sometimes incorrect)
> in spearfish/UTM the intermittend white lines still appear.
>
> r.mapcalc:
> couldn't reproduce the behaviour Anantha Prasad reported.
> This BUG is gone.
> The problem with the reduce/reduce conflicts on compiling
> can IMHO be ignored, as the output of the parser is parsed again
> with a C program. To my knowledge the parser is ok and
> i know of no errors.
> This is the same with r3.mapcalc.
>
> r.in.png does not work
> (r.out.png, then r.in.png with this image gives me a totally white
> raster on IRIX). r.stats says: 100% * (NULL)
>
> r.in.tiff on IRIX:
> r.in.tiff [1591] - out of logical swap space during exec - see swap(1M)
> killed.
> argh, i have 128 MB RAM and another 128 M RAM swap.
> Must be some sort of segfault. This happens when invoking only
> r.in.tiff, without any arguments!
>
>
> wrong order of libs: $(MATHLIB) in
> src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile
>
> src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile
>
>
> CYGWIN:
> grass5
> /usr/local/grass5/etc/Init.sh: not found
> does not start with and without any commandline flag (-text, -tcltk etc.)
>
> COMPLETELY BROKEN NOW!
>
> GENERAL:
>
> g.help:
> PAGEUP/PAGEDOWN: doesn't work on every terminal.
> E. g. on IRIX it is broken.
> This needs to take the termcap into account. It is not possible
> to hardcode the keycodes into the program.
>
> t.roads.inf from spearfish database is broken/wrong format.
> no cat support (informix db?)
>
> tests that i performed:
> for f in `g.mlist type=rast` ; do
> d.erase white
> echo "plotting $f"
> d.rast $f
> sleep 2
> done
> and with d.vect, d.area, d.site etc.
>
> --
> andreas lange, 12/2000

--
Dipl.-Geogr. Markus Neteler * University of Hannover
Institute of Physical Geography and Landscape Ecology
Schneiderberg 50 * D-30167 Hannover * Germany
Tel: ++49-(0)511-762-4494 Fax: -3984

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

6

On Sun, Dec 10, 2000 at 06:30:46AM -0700, John Huddleston wrote:

No Markus, you do not have to change 60% of the Gmakefiles.
if beta10 comes from the cvs.

I modified all the files to be $(LIBS) $(XDRLIB) $(MATHLIB)
when I first started. It is usually new files that potentially may
have incorrect order. The rules depends on which LIBS are
included. Some require XDRLIB and some don't.

O.k.... Do you have any hints *when* XDRLIB is required? How
to find out (another way as bothering Andreas or you :slight_smile:
when to added this to CC line? I'll put a note into SUBMITTING
then (we already have a nice Gmakefile section there!).

Markus

John Huddleston

----- Original Message -----
From: "Markus Neteler" <neteler@geog.uni-hannover.de>
To: <grass5@geog.uni-hannover.de>
Sent: Saturday, December 09, 2000 11:06 AM
Subject: Re: [GRASS5] compiling beta10 on IRIX and CYGWIN/WIN32

> On Sat, Dec 09, 2000 at 01:07:58PM +0100, Andreas Lange wrote:
> > Hi 2 all,
> >
> > i want to report some problems i found with GRASS5.0beta10 (not the CVS
> > version!) on IRIX 6.5 and CYGWIN.
> >
> > I am tired of fixing the same problems again and again. So i want to ask
> > everyone to remember:
> > The Gmakefile lib order _must_ always be:
> > $(LIBS) $(XDRLIB) $(MATHLIB)
> > Otherwise on IRIX it won't compile. If you are linking to the gis lib,
> > you must always include $(MATHLIB), as gis references floor, ceil, etc.
> > from the math lib. If you are not sure, add $(XDRLIB) $(MAHTLIB), it
> > won't hurt.
> Hi Andreas,
>
> actually I don't see the problem:
>
> Is the order:
> $(LIBS) $(MATHLIB) $(XDRLIB)
>
> causing problems? If yes, we have to change 60% (or more) of the Gmakefiles.
> If so, please modify SUBMITTING. I have a script here which could do
> the change of order.
>
> Generally I was sure that 99% of the Gmakefiles are ok. Obviously I am
> wrong?
>
> How does Justin manage if not (he is using Irix, too)?
>
> As Linux doesn't complain if $(XDRLIB) is missing, what is the rule for this
> requirement? Then I can check all Gmakefiles...
>
> Little confused (as beta10 is rather identical to CVS at time),
>
> Markus
>
> PS: I'll look into problems mentioned below:
>
> > I had to heavily edit the header file for CYGWIN, there are still a lot
> > of problems with include and lib search paths etc.
> > And the command line/text mode interface is not working any more.
> >
> > Please see the attached file for the problems in detail.
> >
> > I have currently no time to work on the fixes myself. I will concentrate
> > on the XDRIVER/fifo/ipc stuff.
> >
> > Andreas
> > --
> > Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
> > Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net
>
> > GRASS5.0beta10:
> >
> > IRIX:
> > g.version has double \n's (fix is on the way)
> >
> > gis_set.tcl: -state disabled instead of -state disable
> > gis_set.tcl: string compare -nocase: not supported
> > although this is supported from tk8.2 on, it is not
> > ok to use this if we claim tk8.0 compatibility.
> >
> > when converting the man pages:
> > many error messages with awf formatter:
> > . DE not available (or such)
> >
> > g.manual from tcltkgrass:
> > Illegal option -- P
> > Usage: man ...
> > g.manual entries=d.area -f:
> > Usage:
> > g.manual .....
> >
> > this is all messed up, rman: please do not ask the user to install
> > something from within Usage info. This will confuse "ordinary" users.
> >
> > CELL driver seems to work on IRIX in latlon. (colors sometimes incorrect)
> > in spearfish/UTM the intermittend white lines still appear.
> >
> > r.mapcalc:
> > couldn't reproduce the behaviour Anantha Prasad reported.
> > This BUG is gone.
> > The problem with the reduce/reduce conflicts on compiling
> > can IMHO be ignored, as the output of the parser is parsed again
> > with a C program. To my knowledge the parser is ok and
> > i know of no errors.
> > This is the same with r3.mapcalc.
> >
> > r.in.png does not work
> > (r.out.png, then r.in.png with this image gives me a totally white
> > raster on IRIX). r.stats says: 100% * (NULL)
> >
> > r.in.tiff on IRIX:
> > r.in.tiff [1591] - out of logical swap space during exec - see swap(1M)
> > killed.
> > argh, i have 128 MB RAM and another 128 M RAM swap.
> > Must be some sort of segfault. This happens when invoking only
> > r.in.tiff, without any arguments!
> >
> >
> > wrong order of libs: $(MATHLIB) in
> > src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile
> >
> > src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile
> >
> >
> > CYGWIN:
> > grass5
> > /usr/local/grass5/etc/Init.sh: not found
> > does not start with and without any commandline flag (-text, -tcltk etc.)
> >
> > COMPLETELY BROKEN NOW!
> >
> > GENERAL:
> >
> > g.help:
> > PAGEUP/PAGEDOWN: doesn't work on every terminal.
> > E. g. on IRIX it is broken.
> > This needs to take the termcap into account. It is not possible
> > to hardcode the keycodes into the program.
> >
> > t.roads.inf from spearfish database is broken/wrong format.
> > no cat support (informix db?)
> >
> > tests that i performed:
> > for f in `g.mlist type=rast` ; do
> > d.erase white
> > echo "plotting $f"
> > d.rast $f
> > sleep 2
> > done
> > and with d.vect, d.area, d.site etc.
> >
> > --
> > andreas lange, 12/2000
>
>
> --
> Dipl.-Geogr. Markus Neteler * University of Hannover
> Institute of Physical Geography and Landscape Ecology
> Schneiderberg 50 * D-30167 Hannover * Germany
> Tel: ++49-(0)511-762-4494 Fax: -3984
>
> ----------------------------------------
> If you want to unsubscribe from GRASS Development Team mailing list write to:
> minordomo@geog.uni-hannover.de with
> subject 'unsubscribe grass5'

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

--
Dipl.-Geogr. Markus Neteler * University of Hannover
Institute of Physical Geography and Landscape Ecology
Schneiderberg 50 * D-30167 Hannover * Germany
Tel: ++49-(0)511-762-4494 Fax: -3984

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

7

Hi Markus,

generally all modules/libraries using the external data representation
need the xdrlib. But that should be all modules that link with the gis,
raster and vector library (pretty most of the grass modules!). But as we
found out lateley on Linux and IRIX (and possibly other systems too),
the functions are included in the standard c library. So on most systems
(except Win/Cygwin) there is no need. But the inclusion of $(XDRLIB)
does not hurt in any case.

On IRIX all modules linking with the gis lib and referencing some
function from there that needs a math function (e. g. floor, ceil, sin,
cos etc.) need $(MATHLIB). So too, it will only generate a linker
warning "-lm not used to resolve any function ...".

And the IRIX linker is somewhat confusing as the order is important:
e. g. $(MATHLIB) $(GISLIB) will not work to resolve floor from a
function in the gis lib. $(GISLIB) $(MATHLIB) will work fine. This is a
problem of the IRIX linker used with gcc, so it does not happen on
Linux. Using the MIPS PRO compiler may not have such problem.

I would propose to add $(XDRLIB) $(MATHLIB) always, at least for new
modules.

cu,

Andreas

Markus Neteler wrote:

On Sun, Dec 10, 2000 at 06:30:46AM -0700, John Huddleston wrote:
> No Markus, you do not have to change 60% of the Gmakefiles.
> if beta10 comes from the cvs.
>
> I modified all the files to be $(LIBS) $(XDRLIB) $(MATHLIB)
> when I first started. It is usually new files that potentially may
> have incorrect order. The rules depends on which LIBS are
> included. Some require XDRLIB and some don't.
O.k.... Do you have any hints *when* XDRLIB is required? How
to find out (another way as bothering Andreas or you :slight_smile:
when to added this to CC line? I'll put a note into SUBMITTING
then (we already have a nice Gmakefile section there!).

Markus

> John Huddleston
>

--
Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

8

On Sun, Dec 10, 2000 at 05:12:37PM +0000, Markus Neteler wrote:

On Sun, Dec 10, 2000 at 06:30:46AM -0700, John Huddleston wrote:
> No Markus, you do not have to change 60% of the Gmakefiles.
> if beta10 comes from the cvs.
>
> I modified all the files to be $(LIBS) $(XDRLIB) $(MATHLIB)
> when I first started. It is usually new files that potentially may
> have incorrect order. The rules depends on which LIBS are
> included. Some require XDRLIB and some don't.
O.k.... Do you have any hints *when* XDRLIB is required? How
to find out (another way as bothering Andreas or you :slight_smile:
when to added this to CC line? I'll put a note into SUBMITTING
then (we already have a nice Gmakefile section there!).

Any module that does raster manipulations (from gislib) needs the
XDRLIB.

--
Eric G. Miller <egm2@jps.net>

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

9

On Sun, Dec 10, 2000 at 02:10:20AM +0100, Andreas Lange wrote:

Hi Markus, hi others,

sorry for the confusing message.
The problem is with the two Gmakefiles mentioned later in the text.
Only these two Gmakefiles cause problems because of missing $(MATHLIB)
_or_ $(MATHLIB) _before_ $(GISLIB), which makes the IRIX linker fail.
So the order should be:
$(LIBES) $(WHATEVER) $(GISLIB) $(MATHLIB) $(XDRLIB)

Hi Andreas, hi others,

what about a general rule like this:

LIBES=$(all GRASS GIS libes go here)

$(CC) ... $(LIBES) $(non GRASS libs) $(MATHLIB) $(XDRLIB)

Is this a rule we could add to SUBMITTING?

I just was a little bit frustrated because both programs compiled with
beta8. So everyone should remember not to delete/reorder the LIBES. Then
the Gmakefiles should be 100% correct.

This must have been a mistake as (at least me) most should be aware of this
problem.

The $(XDRLIB) may be useless on linux, but is needed for win32.

I am too overworked right now to look into the Cygwin problem with the
Init.sh. This should be fixed, as the windows port will not work on
commandline :-(. No graphics is bad, but setting the GRASS environment
by hand is nearly impossible.

Tomorrow Justin is back, maybe he has an idea!

Markus

-----------------------
> PS: I'll look into problems mentioned below:
>

> > wrong order of libs: $(MATHLIB) in
> > src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile
> >
> > src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile

--
Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

--
Dipl.-Geogr. Markus Neteler * University of Hannover
Institute of Physical Geography and Landscape Ecology
Schneiderberg 50 * D-30167 Hannover * Germany
Tel: ++49-(0)511-762-4494 Fax: -3984

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

10

Hi Andreas,

On Sun, Dec 10, 2000 at 02:10:20AM +0100, Andreas Lange wrote:

> > wrong order of libs: $(MATHLIB) in
> > src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile
> >
> > src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile

I have fixed both. Please give it a try.
SUBMITTING is updated as well.

Cheers

Markus

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

11

On Sat, Dec 09, 2000 at 01:07:58PM +0100, Andreas Lange wrote:
[...]

GRASS5.0beta10:

IRIX:
g.version has double \n's (fix is on the way)

O.k.

gis_set.tcl: -state disabled instead of -state disable

Can't see that. Seems to be fixed.

gis_set.tcl: string compare -nocase: not supported
although this is supported from tk8.2 on, it is not
ok to use this if we claim tk8.0 compatibility.

This might be done by Justin (hi!).

when converting the man pages:
many error messages with awf formatter:
. DE not available (or such)

Hi Michel?

g.manual from tcltkgrass:
Illegal option -- P
Usage: man ...
g.manual entries=d.area -f:
Usage:
g.manual .....

As g.manual is a script now, the *order* of parameters will be
important. Two possible solutions:

1. change g.manual to accept the flags at *any* place (hi Huidae!)
2. change tcltkgrass/main/gui.tcl to add flags in correct place

this is all messed up, rman: please do not ask the user to install
something from within Usage info. This will confuse "ordinary" users.

CELL driver seems to work on IRIX in latlon. (colors sometimes incorrect)
in spearfish/UTM the intermittend white lines still appear.

r.mapcalc:
couldn't reproduce the behaviour Anantha Prasad reported.
This BUG is gone.

Removed from BUGS.

The problem with the reduce/reduce conflicts on compiling
can IMHO be ignored, as the output of the parser is parsed again
with a C program. To my knowledge the parser is ok and
i know of no errors.
This is the same with r3.mapcalc.

Removed from BUGS.

r.in.png does not work
(r.out.png, then r.in.png with this image gives me a totally white
raster on IRIX). r.stats says: 100% * (NULL)

Maybe we just concentrate on r.in.gdal and add the GDALlib to GRASS sources.

r.in.tiff on IRIX:
r.in.tiff [1591] - out of logical swap space during exec - see swap(1M)
killed.
argh, i have 128 MB RAM and another 128 M RAM swap.
Must be some sort of segfault. This happens when invoking only
r.in.tiff, without any arguments!

As far as I know Eric is working on that.

wrong order of libs: $(MATHLIB) in
src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile

fixed.

src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile

Fixed.

CYGWIN:
grass5
/usr/local/grass5/etc/Init.sh: not found
does not start with and without any commandline flag (-text, -tcltk etc.)
COMPLETELY BROKEN NOW!

Hi Justin! :slight_smile: Seems to be a problem in src/general/init/Gmakefile.

GENERAL:

g.help:
PAGEUP/PAGEDOWN: doesn't work on every terminal.
E. g. on IRIX it is broken.
This needs to take the termcap into account. It is not possible
to hardcode the keycodes into the program.

Volunteers? I have no idea about termcap, sorry.

t.roads.inf from spearfish database is broken/wrong format.
no cat support (informix db?)

Maybe we remove this from spearfish? It will be informix related.

tests that i performed:
for f in `g.mlist type=rast` ; do
  d.erase white
  echo "plotting $f"
  d.rast $f
  sleep 2
done
and with d.vect, d.area, d.site etc.

Sorry, can't follow. You tested all display modules by that?

Regards

Markus

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

12

On Sat, Dec 09, 2000 at 07:26:38PM +0100, Federico Ponchio wrote:

Hi,
  i had the following error message when compiling
the beta 10:

[root@snarf vask]# gmake5
  SRC = /usr/local/src/grass5.0beta10/src
  CMD = /usr/local/src/grass5.0beta10/src/CMD
  HEADER = head.i686-pc-linux-gnu
  ARCH = i686-pc-linux-gnu
  GISBASE = /usr/local/grass5
  VERSION = 5.0beta10 December 2000
#################################################################
/usr/local/src/grass5.0beta10/src/libes/vask
  make -f OBJ.i686-pc-linux-gnu/make.rules gcc -g -O2 -DMAKE_VASK_LIB
-I/usr/include/ncurses.h -I/usr/local/src/grass5.0be ta10/src/include -c
V_call.c -o OBJ.i686-pc-linux-gnu/V_call.o
/usr/include/ncurses.h: Not a directory

which comes from the line:

EXTRA_CFLAGS = -DMAKE_VASK_LIB -I/usr/include/ncurses.h

in src/libes/vask/Gmakefile

I removed the "/ncurses.h" part and all works correctly now.

Hi Federico,

thanks for this! I have removed the hard-coded entry in CVS.

Markus

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

13

Hi all

Markus Neteler wrote:

> gis_set.tcl: -state disabled instead of -state disable
Can't see that. Seems to be fixed.

Yes fixed in CVS

> gis_set.tcl: string compare -nocase: not supported
> although this is supported from tk8.2 on, it is not
> ok to use this if we claim tk8.0 compatibility.
This might be done by Justin (hi!).

Agian, fixed in CVS

> CYGWIN:
> grass5
> /usr/local/grass5/etc/Init.sh: not found
> does not start with and without any commandline flag (-text, -tcltk
> etc.)
> COMPLETELY BROKEN NOW!
Hi Justin! :slight_smile: Seems to be a problem in src/general/init/Gmakefile.

I need more information for this. /usr/local/grass5/etc/Init.sh seems
to be the correct path (at least for Unix). What should it be for
CYGWIN?

--
Sincerely,

Jazzman (a.k.a. Justin Hickey) e-mail: jhickey@hpcc.nectec.or.th
High Performance Computing Center
National Electronics and Computer Technology Center (NECTEC)
Bangkok, Thailand

People who think they know everything are very irritating to those
of us who do. ---Anonymous

Jazz and Trek Rule!!!

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

14

Hi Markus,

thank you very much for inspecting the errors despite the partly
confusing text.

Markus Neteler wrote:

> when converting the man pages:
> many error messages with awf formatter:
> . DE not available (or such)
Hi Michel?

I fear that this is a problem of the IRIX awf formatter. I'll have a
look at the man pages another time.

> g.manual from tcltkgrass:
> Illegal option -- P
> Usage: man ...
> g.manual entries=d.area -f:
> Usage:
> g.manual .....
As g.manual is a script now, the *order* of parameters will be
important. Two possible solutions:

Hmm. It was my impression that the IRIX man has no -P option what caused
the above error.

1. change g.manual to accept the flags at *any* place (hi Huidae!)
2. change tcltkgrass/main/gui.tcl to add flags in correct place

> this is all messed up, rman: please do not ask the user to install
> something from within Usage info. This will confuse "ordinary" users.
>

> wrong order of libs: $(MATHLIB) in
> src.garden/answers/src.answers/display/d.lingraph/cmd/Gmakefile
fixed.

> src.contrib/GMSL/g3d/src3d/raster/r3.showdspf.openGL/Gmakefile
Fixed.

Thank you for fixing this.

> CYGWIN:
> grass5
> /usr/local/grass5/etc/Init.sh: not found
> does not start with and without any commandline flag (-text, -tcltk etc.)
> COMPLETELY BROKEN NOW!
Hi Justin! :slight_smile: Seems to be a problem in src/general/init/Gmakefile.

> GENERAL:
>
> g.help:
> PAGEUP/PAGEDOWN: doesn't work on every terminal.
> E. g. on IRIX it is broken.
> This needs to take the termcap into account. It is not possible
> to hardcode the keycodes into the program.
Volunteers? I have no idea about termcap, sorry.

> t.roads.inf from spearfish database is broken/wrong format.
> no cat support (informix db?)
Maybe we remove this from spearfish? It will be informix related.

Yes i would propose to remove it sometimes. But nothing with priority.

> tests that i performed:
> for f in `g.mlist type=rast` ; do
> d.erase white
> echo "plotting $f"
> d.rast $f
> sleep 2
> done
> and with d.vect, d.area, d.site etc.
Sorry, can't follow. You tested all display modules by that?

I did only a visual test of all spearfish raster/vector files this way.
I hope to get my expect/dejagnu setup up again before christmas.

cu,

Andreas
--
Andreas Lange, 65187 Wiesbaden, Germany, Tel. +49 611 807850
Andreas.Lange@Rhein-Main.de - A.C.Lange@GMX.net

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

15

Markus Neteler wrote:

On Sat, Dec 09, 2000 at 07:26:38PM +0100, Federico Ponchio wrote:
> Hi,
> i had the following error message when compiling
> the beta 10:
>
> [root@snarf vask]# gmake5
> SRC = /usr/local/src/grass5.0beta10/src
> CMD = /usr/local/src/grass5.0beta10/src/CMD
> HEADER = head.i686-pc-linux-gnu
> ARCH = i686-pc-linux-gnu
> GISBASE = /usr/local/grass5
> VERSION = 5.0beta10 December 2000
> #################################################################
> /usr/local/src/grass5.0beta10/src/libes/vask
> make -f OBJ.i686-pc-linux-gnu/make.rules gcc -g -O2 -DMAKE_VASK_LIB
> -I/usr/include/ncurses.h -I/usr/local/src/grass5.0be ta10/src/include -c
> V_call.c -o OBJ.i686-pc-linux-gnu/V_call.o
> /usr/include/ncurses.h: Not a directory
>
> which comes from the line:
>
> EXTRA_CFLAGS = -DMAKE_VASK_LIB -I/usr/include/ncurses.h
>
> in src/libes/vask/Gmakefile
>
> I removed the "/ncurses.h" part and all works correctly now.

Hi Federico,

thanks for this! I have removed the hard-coded entry in CVS.

Markus

----------------------------------------
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo@geog.uni-hannover.de with
subject 'unsubscribe grass5'

In my system i have a dir called ncruses as well as having ncurses.h in
the parent dir.
I have noticed that the up, down, left, right, del, backspace or space.
which is a hassle when you have to set up a new location and especially
a default region. thanks for the other info i got grass to compile fine
with the exception of png stuff and clump...