[GRASS5] Win32 compile with cygnus

GRASS GRASS developer
1

Hi,

here now the commented error.log of my first attempt to compile on
Win32.

That looks quite good, only some minor problems. Currently i have not
installed the X libraries, i expect to arrive a complete compiler/tools
CD next week.

Any idea what the problem with the fonts is?

I'll try to fix the other problems, if someone knows an apparent
solution, please tell me so that i don't waste time on something that
has a known solution.

cu,

Andreas

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

(attachments)

error.log (1.59 KB)

2

Hi Andreas,

thanks for testing GRASS on Cygwin!

Here some comments:

On Sun, Nov 19, 2000 at 06:10:16PM +0100, Andreas Lange wrote:

Compilation error in module: html (ignored)
- perl missing

-> Hi Michel, hi all, is anybody willing to change g.html2man from
   PERL to shell script? It's the only PERL dependency, as far as I know.

Compilation error in module: src/raster/r.agnps50 (ignored)
- missing $(XDRLIB) in r.cn/Gmakefile

Fixed.

- sys/config.h permission denied ???

?? Looking at the #include statements, it looking normal.

Compilation error in module: src/raster/r.in.gdal (ignored)
- gbgetsymbol.c:111: redefinition of GBGetSymbol

Hi Frank W. :slight_smile:

Compilation error in module: src/raster/r.in.png (ignored)
- pnm.h not found in pgnfunc.h
- libpng only knows png.h and pngconf.h ?

It seems to require the "netpbm" lib.
I have it here:
/usr/X11R6/include/pnm.h
It would be good to remove the netpbm dependency (TODO).

Compilation error in module: src/raster/r.out.png (ignored)
- pnm.h, ppm.h missing

see r.in.png comment

Compilation error in module: src/raster/r.random (ignored)
FLT_MAX, FLT_MIN, DBL_MIN undeclared

I have added
#include <float.h>
into count.c. Please try again.

Compilation error in module: src/sites/s.windavg (ignored)
- missing xdr lib in Gmakefile

fixed.

Compilation error in module: src.contrib/CERL/raster/r.linear.regression (ignored)
- solve.c:14: multiple definition of solve

I have renamed solve to solve2 in nlfit.c
Please try again.

Compilation error in module: src.contrib/CERL/raster/r.rational.regression (ignored)
- nlfit.c: ... multiple definition of solve

I have renamed solve to solve2 in nlfit.c
Please try again.

Hope this help a bit

Markus

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

3

Hi Markus,

your are much too fast for me ;-). I want to first check any changes on
Linux and with the cygnus setup, but re-compiling with cygnus needs
several hours. Now i know what John Huddleston is talking about.

The problem with the pnm.h etc is indeed that the r.in.png and r.out.png
depend on the netpbm library. This is always installed on Linux and most
other Unix. I never tested if this library compiles under Cygwin. I'll
add this dependency to the REQUIREMENTS file.

I personally think that we shouldn't waste time on re-writing
r.in.png/r.out.png to remove the dependency on this library, but look
for a more generic solution for the import of non-georeferenced images.
I'll check in some thoughts of mine in the next days. I think we should
look for a general library that handles the usual image formats (png,
jpeg, etc.).

For the float.h: does this introduce problems on other systems?
maybe a #ifdef __CYGWIN__ would be better.

For g.html2man:
Another solution would be to maintain the man-pages on the server and to
automatically update them there and check them in. I have no idea if
this is possible or better.
Converting the perl script to a sed/shell script is possible, but this
would hurt performance (we had this situation already!).

Does anybody know if there is a tcl/tk X11 wish with the Cygwin port of
X11R6? If not, i see not how tcltkgrass will work in the cygwin port.

cu,

Andreas

Markus Neteler wrote:

Hi Andreas,

thanks for testing GRASS on Cygwin!

Here some comments:

On Sun, Nov 19, 2000 at 06:10:16PM +0100, Andreas Lange wrote:
> Compilation error in module: html (ignored)
> - perl missing

-> Hi Michel, hi all, is anybody willing to change g.html2man from
   PERL to shell script? It's the only PERL dependency, as far as I know.

> Compilation error in module: src/raster/r.agnps50 (ignored)
> - missing $(XDRLIB) in r.cn/Gmakefile
Fixed.

> - sys/config.h permission denied ???
?? Looking at the #include statements, it looking normal.

> Compilation error in module: src/raster/r.in.gdal (ignored)
> - gbgetsymbol.c:111: redefinition of GBGetSymbol
Hi Frank W. :slight_smile:

> Compilation error in module: src/raster/r.in.png (ignored)
> - pnm.h not found in pgnfunc.h
> - libpng only knows png.h and pngconf.h ?
It seems to require the "netpbm" lib.
I have it here:
/usr/X11R6/include/pnm.h
It would be good to remove the netpbm dependency (TODO).

> Compilation error in module: src/raster/r.out.png (ignored)
> - pnm.h, ppm.h missing
see r.in.png comment

> Compilation error in module: src/raster/r.random (ignored)
> FLT_MAX, FLT_MIN, DBL_MIN undeclared
I have added
#include <float.h>
into count.c. Please try again.

> Compilation error in module: src/sites/s.windavg (ignored)
> - missing xdr lib in Gmakefile
fixed.

> Compilation error in module: src.contrib/CERL/raster/r.linear.regression (ignored)
> - solve.c:14: multiple definition of solve
I have renamed solve to solve2 in nlfit.c
Please try again.

> Compilation error in module: src.contrib/CERL/raster/r.rational.regression (ignored)
> - nlfit.c: ... multiple definition of solve
I have renamed solve to solve2 in nlfit.c
Please try again.

Hope this help a bit

Markus

--
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'

4

Hi Andreas,

On Sun, Nov 19, 2000 at 08:18:21PM +0100, Andreas Lange wrote:

Hi Markus,

your are much too fast for me ;-).

Just wanted to be a bit of help :slight_smile:

I want to first check any changes on
Linux and with the cygnus setup, but re-compiling with cygnus needs
several hours. Now i know what John Huddleston is talking about.

I see, I didn't know that.

The problem with the pnm.h etc is indeed that the r.in.png and r.out.png
depend on the netpbm library. This is always installed on Linux and most
other Unix. I never tested if this library compiles under Cygwin. I'll
add this dependency to the REQUIREMENTS file.

O.k.

I personally think that we shouldn't waste time on re-writing
r.in.png/r.out.png to remove the dependency on this library, but look
for a more generic solution for the import of non-georeferenced images.
I'll check in some thoughts of mine in the next days. I think we should
look for a general library that handles the usual image formats (png,
jpeg, etc.).

Yes, that would be the best. Probably the GDAL could help here.

For the float.h: does this introduce problems on other systems?
maybe a #ifdef __CYGWIN__ would be better.

I don't think this ifdef is required. I had to add it to r.in.dem as
well (for MacOS). The Linux likes it, too. It's present on SUN/Solaris.
On CRAY it is o.k., too. What about SGI?

For g.html2man:
Another solution would be to maintain the man-pages on the server and to
automatically update them there and check them in. I have no idea if
this is possible or better.

Mhh, I don't get it. You think to maintain MAN format? At least I
don't know how to write it (HTML is much easier). If you think of an
*online* version, please consider that many users don't have
permanent access (or access at all, those CDROM users).

Converting the perl script to a sed/shell script is possible, but this
would hurt performance (we had this situation already!).

Yes, probably we need PERL as requirement...

Does anybody know if there is a tcl/tk X11 wish with the Cygwin port of
X11R6? If not, i see not how tcltkgrass will work in the cygwin port.

Hi John! Any suggestions?

Andreas, did you see this:
http://www.scriptics.com/customers/success/cygnus.tml?sc_format=print
I don't know if it is the right thing.

Regards

Markus

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

5

On Sun, Nov 19, 2000 at 08:05:59PM +0000, Markus Neteler wrote:

> I personally think that we shouldn't waste time on re-writing
> r.in.png/r.out.png to remove the dependency on this library, but look
> for a more generic solution for the import of non-georeferenced images.
> I'll check in some thoughts of mine in the next days. I think we should
> look for a general library that handles the usual image formats (png,
> jpeg, etc.).
Yes, that would be the best. Probably the GDAL could help here.

I'd think netpbm is pretty portable. But I could be wrong.

> For the float.h: does this introduce problems on other systems?
> maybe a #ifdef __CYGWIN__ would be better.
I don't think this ifdef is required. I had to add it to r.in.dem as
well (for MacOS). The Linux likes it, too. It's present on SUN/Solaris.
On CRAY it is o.k., too. What about SGI?

What's the problem with float.h? I know I've used it in a couple places
for DBL_MIN, DBL_MAX and friends (not in limits.h). I thought this was
a standard header ???

--
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'

6

Hi Eric,

Eric G . Miller wrote:

On Sun, Nov 19, 2000 at 08:05:59PM +0000, Markus Neteler wrote:
> > I personally think that we shouldn't waste time on re-writing
> > r.in.png/r.out.png to remove the dependency on this library, but look
> > for a more generic solution for the import of non-georeferenced images.
> > I'll check in some thoughts of mine in the next days. I think we should
> > look for a general library that handles the usual image formats (png,
> > jpeg, etc.).
> Yes, that would be the best. Probably the GDAL could help here.

I'd think netpbm is pretty portable. But I could be wrong.

Yes, netpbm is portable. There is IMHO even a DOS version. But i think
that it would be better in the long run to have _one_ import program for
all image formats instead for the r.in.{png|gif|tiff|...} and
r.out.{png|gif|tiff|...}. This is much work to maintain and confusing
for users.

> > For the float.h: does this introduce problems on other systems?
> > maybe a #ifdef __CYGWIN__ would be better.
> I don't think this ifdef is required. I had to add it to r.in.dem as
> well (for MacOS). The Linux likes it, too. It's present on SUN/Solaris.
> On CRAY it is o.k., too. What about SGI?

What's the problem with float.h? I know I've used it in a couple places
for DBL_MIN, DBL_MAX and friends (not in limits.h). I thought this was
a standard header ???

We discussed this some time ago on this list. But i can't memorize the
result. I faintly remember problems with the BSD flavours. There must be
a reason why Huidae Cho didn't include float.h i assume.

A side note to Markus: Please refer to Mac OS X, not to MacOS, as that
are two _very_ different things.

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'

7

On Sun, Nov 19, 2000 at 08:18:21PM +0100, Andreas Lange wrote:

The problem with the pnm.h etc is indeed that the r.in.png and r.out.png
depend on the netpbm library. This is always installed on Linux and most
other Unix.

Not so. On many Linux distributions it is not installed unless some
other product depends on it. It is fairly small (528Kb Intel bin) and
so placing a dependency on it does not seem to me to be a worry.

I personally think that we shouldn't waste time on re-writing
r.in.png/r.out.png to remove the dependency on this library, but look
for a more generic solution for the import of non-georeferenced images.
I'll check in some thoughts of mine in the next days. I think we should
look for a general library that handles the usual image formats (png,
jpeg, etc.).

Last I checked, netpbm is the best for this sort of stuff. Rather than
write a converter from any image format to any other, netpbm solves the
issue by adopting a common intermediate format, and supplying converters
to and from that format.

--
James Cameron mailto:quozl@us.netrek.org http://quozl.netrek.org/

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

8

Andreas Lange wrote:

Eric G . Miller wrote:
> > > For the float.h: does this introduce problems on other systems?
> > > maybe a #ifdef __CYGWIN__ would be better.
> > I don't think this ifdef is required. I had to add it to r.in.dem
> > as well (for MacOS). The Linux likes it, too. It's present on
> > SUN/Solaris. On CRAY it is o.k., too. What about SGI?
>
> What's the problem with float.h? I know I've used it in a couple
> places for DBL_MIN, DBL_MAX and friends (not in limits.h). I
> thought this was a standard header ???

Yes, it is a standard header, since it appears in the standard library
listed in Appendix B of "The C Programming Language" by Kernighan and
Ritchie. There should be no problems using it.

We discussed this some time ago on this list. But i can't memorize the
result. I faintly remember problems with the BSD flavours. There must
be a reason why Huidae Cho didn't include float.h i assume.

I don't recall this discussion, can anyone remember what the problem
was? Is BSD using a non-standard float.h file?

--
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'

9

Markus Neteler wrote:

Hi Andreas,

thanks for testing GRASS on Cygwin!

Here some comments:

On Sun, Nov 19, 2000 at 06:10:16PM +0100, Andreas Lange wrote:
> Compilation error in module: html (ignored)
> - perl missing

-> Hi Michel, hi all, is anybody willing to change g.html2man from
   PERL to shell script? It's the only PERL dependency, as far as I know.

A shell script will use awk and sed. I don't know if they are available
on WINxx... Or write some lex+yacc parser... this will only be usefull if
the man command exists on WINxx !

--
Michel WURTZ - DIG - Maison de la télédétection
               500, rue J.F. Breton
               34093 MONTPELLIER Cedex 5

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