octave-dev — GNU Octave developers' list

[OctDev] Correct czt implementation

[<dg...@um...>]

[Apologies for the double post, but I forgot the subject line the first tim=
e!]

The czt function was not working, so I implemented it correctly.  At a
minimum, a correct czt should pass this test:

x=3Drand(1000,1);
y1=3Dfft(x);
y2=3Dczt(x);
max(abs(y1-y2)) # should be a very small number

Also fixed the documentation.

Will someone put this in the cvs?  Thanks.

- DG

## Copyright (C) 2004 Daniel Gunyan
##
## This program is free software; you can redistribute it and/or modify
## it under the terms of the GNU General Public License as published by
## the Free Software Foundation; either version 2 of the License, or
## (at your option) any later version.
##
## This program is distributed in the hope that it will be useful,
## but WITHOUT ANY WARRANTY; without even the implied warranty of
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
## GNU General Public License for more details.
##
## You should have received a copy of the GNU General Public License
## along with this program; if not, write to the Free Software
## Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA

## usage y=3Dczt(x, m, w, a)
##
## Chirp z-transform.  Compute the frequency response starting at a and
## stepping by w for m steps.  a is a point in the complex plane, and
## w is the ratio between points in each step (i.e., radius increases
## exponentially, and angle increases linearly).
##
## To evaluate the frequency response for the range f1 to f2 in a signal
## with sampling frequency Fs, use the following:
##     m =3D 32;                               ## number of points desired
##     w =3D exp(-j*2*pi*(f2-f1)/((m-1)*Fs));  ## freq. step of f2-f1/m
##     a =3D exp(j*2*pi*f1/Fs);                ## starting at frequency f1
##     y =3D czt(x, m, w, a);
##
## If you don't specify them, then the parameters default to a fourier=20
## transform:
##     m=3Dlength(x), w=3Dexp(-j*2*pi/m), a=3D1
##
## If x is a matrix, the transform will be performed column-by-column.

## Algorithm (based on Oppenheim and Schafer, "Discrete-Time Signal
## Processing", pp. 623-628):
##   make chirp of length -N+1 to max(N-1,M-1)
##     chirp =3D> w^([-N+1:max(N-1,M-1)]^2/2)
##   multiply x by chirped a and by N-elements of chirp, and call it g
##   convolve g with inverse chirp, and call it gg
##     pad ffts so that multiplication works
##     ifft(fft(g)*fft(1/chirp))
##   multiply gg by M-elements of chirp and call it done

function y =3D czt(x, m, w, a)
  if nargin < 1 || nargin > 4, usage("y=3Dczt(x, m, w, a)"); endif

  [row, col] =3D size(x);
  if row =3D=3D 1, x =3D x(:); col =3D 1; endif

  if nargin < 2 || isempty(m), m =3D length(x(:,1)); endif
  if length(m) > 1, error("czt: m must be a single element\n"); endif
  if nargin < 3 || isempty(w), w =3D exp(-2*j*pi/m); endif
  if nargin < 4 || isempty(a), a =3D 1; endif
  if length(w) > 1, error("czt: w must be a single element\n"); endif
  if length(a) > 1, error("czt: a must be a single element\n"); endif

  ## indexing to make the statements a little more compact
  n =3D length(x(:,1));
  N =3D [0:n-1]'+n;
  NM =3D [-(n-1):(m-1)]'+n;
  M =3D [0:m-1]'+n;

  nfft =3D 2^nextpow2(n+m-1); # fft pad
  W2 =3D w.^(([-(n-1):max(m-1,n-1)]'.^2)/2); # chirp

  for idx =3D 1:col
    fg =3D fft(x(:,idx).*(a.^-(N-n)).*W2(N), nfft);
    fw =3D fft(1./W2(NM), nfft);
    gg =3D ifft(fg.*fw, nfft);

    y(:,idx) =3D gg(M).*W2(M);
  endfor

  if row =3D=3D 1, y =3D y.'; endif
endfunction

--=20
___________________________________________________________
Sign-up for Ads Free at Mail.com
http://promo.mail.com/adsfreejump.htm

[OctDev] (no subject)

[<dg...@um...>]

(It seems the octave-dev list is very quiet, so I'm also sending this to the
octave-help list.)

The czt function was not working, so I implemented it correctly.  At a
minimum, a correct czt should pass this test:

x=3Drand(1000,1);
y1=3Dfft(x);
y2=3Dczt(x);
max(abs(y1-y2)) # should be a very small number

Also fixed the documentation.

Will someone put this in the cvs?  Thanks.

- DG

## Copyright (C) 2004 Daniel Gunyan
##
## This program is free software; you can redistribute it and/or modify
## it under the terms of the GNU General Public License as published by
## the Free Software Foundation; either version 2 of the License, or
## (at your option) any later version.
##
## This program is distributed in the hope that it will be useful,
## but WITHOUT ANY WARRANTY; without even the implied warranty of
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
## GNU General Public License for more details.
##
## You should have received a copy of the GNU General Public License
## along with this program; if not, write to the Free Software
## Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA

## usage y=3Dczt(x, m, w, a)
##
## Chirp z-transform.  Compute the frequency response starting at a and
## stepping by w for m steps.  a is a point in the complex plane, and
## w is the ratio between points in each step (i.e., radius increases
## exponentially, and angle increases linearly).
##
## To evaluate the frequency response for the range f1 to f2 in a signal
## with sampling frequency Fs, use the following:
##     m =3D 32;                               ## number of points desired
##     w =3D exp(-j*2*pi*(f2-f1)/((m-1)*Fs));  ## freq. step of f2-f1/m
##     a =3D exp(j*2*pi*f1/Fs);                ## starting at frequency f1
##     y =3D czt(x, m, w, a);
##
## If you don't specify them, then the parameters default to a fourier=20
## transform:
##     m=3Dlength(x), w=3Dexp(-j*2*pi/m), a=3D1
##
## If x is a matrix, the transform will be performed column-by-column.

## Algorithm (based on Oppenheim and Schafer, "Discrete-Time Signal
## Processing", pp. 623-628):
##   make chirp of length -N+1 to max(N-1,M-1)
##     chirp =3D> w^([-N+1:max(N-1,M-1)]^2/2)
##   multiply x by chirped a and by N-elements of chirp, and call it g
##   convolve g with inverse chirp, and call it gg
##     pad ffts so that multiplication works
##     ifft(fft(g)*fft(1/chirp))
##   multiply gg by M-elements of chirp and call it done

function y =3D czt(x, m, w, a)
  if nargin < 1 || nargin > 4, usage("y=3Dczt(x, m, w, a)"); endif

  [row, col] =3D size(x);
  if row =3D=3D 1, x =3D x(:); col =3D 1; endif

  if nargin < 2 || isempty(m), m =3D length(x(:,1)); endif
  if length(m) > 1, error("czt: m must be a single element\n"); endif
  if nargin < 3 || isempty(w), w =3D exp(-2*j*pi/m); endif
  if nargin < 4 || isempty(a), a =3D 1; endif
  if length(w) > 1, error("czt: w must be a single element\n"); endif
  if length(a) > 1, error("czt: a must be a single element\n"); endif

  ## indexing to make the statements a little more compact
  n =3D length(x(:,1));
  N =3D [0:n-1]'+n;
  NM =3D [-(n-1):(m-1)]'+n;
  M =3D [0:m-1]'+n;

  nfft =3D 2^nextpow2(n+m-1); # fft pad
  W2 =3D w.^(([-(n-1):max(m-1,n-1)]'.^2)/2); # chirp

  for idx =3D 1:col
    fg =3D fft(x(:,idx).*(a.^-(N-n)).*W2(N), nfft);
    fw =3D fft(1./W2(NM), nfft);
    gg =3D ifft(fg.*fw, nfft);

    y(:,idx) =3D gg(M).*W2(M);
  endfor

  if row =3D=3D 1, y =3D y.'; endif
endfunction

--=20
___________________________________________________________
Sign-up for Ads Free at Mail.com
http://promo.mail.com/adsfreejump.htm

[OctDev] Octave wiki

[Etienne G. <et...@cs...>]

  Hi all,

I changed the entry text on the octave wiki.octave.org to

======================================================================
  
  Unfortunately, anyone cannot contribute to this site anymore. If you
  wish to contribute, please send an email to etienne at cs dot uky
  dot eee dee you, saying the IP of the machine from which you wish to
  edit this wiki.

  Hopefully, this is only a temporary fix. This site has been badly
  defaced recently. Thanks to David Bateman for repairing the damage,
  ringing the alarm and proposing fixes. I have little time for
  administration right now, but hope to soon find a solution that
  preserves the "wiki" spirit in this site.

======================================================================

  More news, hopefully, soon.

  Etienne

-- 
Etienne Grossmann ------ http://www.cs.uky.edu/~etienne

[OctDev] irsa_resample patch

[<dg...@um...>]

Will someone add this to the cvs please?  The lm parameter won't work wit=
hout it.

--- /usr/share/octave/2.1.57/site/m/octave-forge/irsa/irsa_resample.m   2=
004-10-07 11:23:03.000000000 -0700
+++ /usr/share/octave/2.1.57/site/m/octave-forge/irsa/irsa_resample.m.ori=
g      2004-07-07 18:42:32.000000000 -0700
@@ -40,7 +40,7 @@

 function newyp =3D irsa_resample (xp, yp, newxp, lm )

-  if( nargin < 3 )
+  if( nargin !=3D 3 )
     usage( "newyp =3D irsa_resample (xp, yp, newxp, [lm] )" );
   endif


--
Daniel


--=20
___________________________________________________________
Sign-up for Ads Free at Mail.com
http://promo.mail.com/adsfreejump.htm

[OctDev] Missing values

[Teemu I. <tpi...@pc...>]

Hi all,

Support for missing double values (NA) was added to Octave two years ago.
There was talk of making functions like sum, mean, etc. to support them, but
e.g. sum([1,NA]) still returns NA. 
(see http://www.octave.org/mailing-lists/help-octave/2002/581)

Also, min([1,NA]) returns 1, but so does min([1,NaN]), which would imply
that NA = NaN in the current implementation. Is this because no one has
fixed the code, or is it so by design?

Octave-forge includes functions nanmax, nanmean etc. which ignore NaN as
they should, but return NA on any vector containing NA. Should they be made
to also ignore NA?

Teemu

Re: [OctDev] Issue Building/Installing liboctave_fixed.dll.a

[David B. <Dav...@mo...>]

Just to round up this issue, the problem was a change in mkoctfile in 
2.1.54 where SH_LDFLAGS was changed to DL_LDFLAGS. Due to this change
the necessary linker flag to build the *.dll.a file wasn't being passed
to the linker..

I've committed a patch, that James confirms fixes the problem for him..

Regards
David

-- 
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

RE: [OctDev] Issue Building/Installing liboctave_fixed.dll.a

[Phillips, J. R <jam...@ho...>]

OK, here are some cygwin command-line queries and results:

$ cd /usr/local
$ find . -name liboctave.dll* -print
./bin/liboctave.dll
./lib/octave-2.1.60/liboctave.dll.a.2.1.60
./lib/octave-2.1.60/liboctave.dll.a
$ find . -name liboctave_fixed* -print
./bin/liboctave_fixed.dll
$ ld --version
GNU ld version 2.15.91 20040725
...
$ g++ --version
g++ (GCC) 3.3.3 (cygwin special)
...

These are the most recent stable versions of g++ and ld available from cygwin.

When using "make -k install", I didn't get appreciably different results than without the "-k" flag.  The install
basically quits copying files once it hits an error, even for subdirs other than main/fixed.  So to finish the install
successfully, it is still necessary to touch main/fixed/NOINSTALL, otherwise a lot of files not in main/fixed don't get
copied.

JRP


-----Original Message-----
From: David Bateman [mailto:Dav...@mo...]
Sent: Wednesday, September 29, 2004 11:57 AM
To: Phillips, James R
Cc: 'David Bateman'; oct...@li...
Subject: Re: [OctDev] Issue Building/Installing liboctave_fixed.dll.a


According to Phillips, James R <jam...@ho...> (on 09/29/04):
> No build errors are generated, but the file liboctave_fixed.dll.a is not made.  This is true even when executing make
in
> the main/fixed subdirectory.  I also explicitly configured octave-forge with the --enable-fixed --disable-static flags
> prior to compiling, but this made no difference.

The *.dll.a is created by ld with the option --out-implib=$(FIXEDLIBTARGET).a
The fact that it isn't created, if there are no build errors, means that the
options of cygwin's ld have changed... Is there a liboctave.dll and a 
liboctave.dll.a for example? Is there even a liboctave_fixed.dll ?
What version of ld and g++ have you installed?

> This file appears to be an export library, which would only be
> required if one wished to link an octave extension (.oct) against
> liboctave_fixed.dll.  Is this correct?  If so, could I force the
> installation to proceed, despite the absence of this file, which is
> not really necessary for use of the toolbox in the octave
> environment?

make -k install should force the install, and would at worst mean that
you couldn't use the fixed-point toolbox....

The last time I booted into windows was more than 6 months ago, and so 
it is difficult for me to debug this.. I suppose I'll have though no
promises of when I'll get on to it.. Can someone else who builds
with cygwin tell me if they get this error?

D.

-- 
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

Re: [OctDev] Issue Building/Installing liboctave_fixed.dll.a

[David B. <Dav...@mo...>]

According to Phillips, James R <jam...@ho...> (on 09/29/04):
> No build errors are generated, but the file liboctave_fixed.dll.a is not made.  This is true even when executing make in
> the main/fixed subdirectory.  I also explicitly configured octave-forge with the --enable-fixed --disable-static flags
> prior to compiling, but this made no difference.

The *.dll.a is created by ld with the option --out-implib=$(FIXEDLIBTARGET).a
The fact that it isn't created, if there are no build errors, means that the
options of cygwin's ld have changed... Is there a liboctave.dll and a 
liboctave.dll.a for example? Is there even a liboctave_fixed.dll ?
What version of ld and g++ have you installed?

> This file appears to be an export library, which would only be
> required if one wished to link an octave extension (.oct) against
> liboctave_fixed.dll.  Is this correct?  If so, could I force the
> installation to proceed, despite the absence of this file, which is
> not really necessary for use of the toolbox in the octave
> environment?

make -k install should force the install, and would at worst mean that
you couldn't use the fixed-point toolbox....

The last time I booted into windows was more than 6 months ago, and so 
it is difficult for me to debug this.. I suppose I'll have though no
promises of when I'll get on to it.. Can someone else who builds
with cygwin tell me if they get this error?

D.

-- 
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

RE: [OctDev] Issue Building/Installing liboctave_fixed.dll.a

[Phillips, J. R <jam...@ho...>]

No build errors are generated, but the file liboctave_fixed.dll.a is not made.  This is true even when executing make in
the main/fixed subdirectory.  I also explicitly configured octave-forge with the --enable-fixed --disable-static flags
prior to compiling, but this made no difference.

This file appears to be an export library, which would only be required if one wished to link an octave extension (.oct)
against liboctave_fixed.dll.  Is this correct?  If so, could I force the installation to proceed, despite the absence of
this file, which is not really necessary for use of the toolbox in the octave environment?

JRP

-----Original Message-----
From: David Bateman [mailto:Dav...@mo...]
Sent: Wednesday, September 29, 2004 11:34 AM
To: Phillips, James R
Cc: oct...@li...
Subject: Re: [OctDev] Issue Building/Installing liboctave_fixed.dll.a


According to Phillips, James R <jam...@ho...> (on 09/29/04):
> Hi,
> 
> I have been trying to build/install octave-2.1.60 together with
> octave-forge-2004.09.09 on cygwin.  Octave-2.1.60 is configured with
> --enable-shared, --disable-static, and compiles and installs successfully.
> The octave-forge configure and compile seem to complete successfully, but
> during installation of main/fixed, the error message
> 
> ****** WARNING: liboctave_fixed.dll.a not correctly installed
> ****** To ensure correct operation liboctave_fixed.dll.a
> ****** should be in the same directory as liboctave.dll.a
> /usr/bin/install: cannot stat `liboctave_fixed.dll.a': No such file or
> directory
> 
> is generated.

Did you get a build error in main/fixed? If so then this might explain the
problem. Note also the Makefile for this is a bit dummy at the moment and
doesn't really try and probe the correct install directory for 
liboctave_fixed.dll.a, so you might have to define "--prefix=" for the
configuration of octave-forge... 

> It appears that the makefile is not generating instructions to build the
> liboctave_fixed.dll.a, while the install is requiring it.  The library
> liboctave_fixed.dll _is_  present in the main/fixed directory.  The only
> workaround I have found is to create main/fixed/NOINSTALL, so the main/fixed
> libraries will not be installed.  
> 
> Since the build seemed to be successful, it seems a shame not to install the
> fixed point tools.  Can anyone point me to a workaround?

If its not built its an error. Try typing "make" directly in main/fixed
and see if it generates any error messages..

D.
-- 
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

Re: [OctDev] Issue Building/Installing liboctave_fixed.dll.a

[David B. <Dav...@mo...>]

According to Phillips, James R <jam...@ho...> (on 09/29/04):
> Hi,
> 
> I have been trying to build/install octave-2.1.60 together with
> octave-forge-2004.09.09 on cygwin.  Octave-2.1.60 is configured with
> --enable-shared, --disable-static, and compiles and installs successfully.
> The octave-forge configure and compile seem to complete successfully, but
> during installation of main/fixed, the error message
> 
> ****** WARNING: liboctave_fixed.dll.a not correctly installed
> ****** To ensure correct operation liboctave_fixed.dll.a
> ****** should be in the same directory as liboctave.dll.a
> /usr/bin/install: cannot stat `liboctave_fixed.dll.a': No such file or
> directory
> 
> is generated.

Did you get a build error in main/fixed? If so then this might explain the
problem. Note also the Makefile for this is a bit dummy at the moment and
doesn't really try and probe the correct install directory for 
liboctave_fixed.dll.a, so you might have to define "--prefix=" for the
configuration of octave-forge... 

> It appears that the makefile is not generating instructions to build the
> liboctave_fixed.dll.a, while the install is requiring it.  The library
> liboctave_fixed.dll _is_  present in the main/fixed directory.  The only
> workaround I have found is to create main/fixed/NOINSTALL, so the main/fixed
> libraries will not be installed.  
> 
> Since the build seemed to be successful, it seems a shame not to install the
> fixed point tools.  Can anyone point me to a workaround?

If its not built its an error. Try typing "make" directly in main/fixed
and see if it generates any error messages..

D.
-- 
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

[OctDev] Issue Building/Installing liboctave_fixed.dll.a

[Phillips, J. R <jam...@ho...>]

Hi,

I have been trying to build/install octave-2.1.60 together with
octave-forge-2004.09.09 on cygwin.  Octave-2.1.60 is configured with
--enable-shared, --disable-static, and compiles and installs successfully.
The octave-forge configure and compile seem to complete successfully, but
during installation of main/fixed, the error message

****** WARNING: liboctave_fixed.dll.a not correctly installed
****** To ensure correct operation liboctave_fixed.dll.a
****** should be in the same directory as liboctave.dll.a
/usr/bin/install: cannot stat `liboctave_fixed.dll.a': No such file or
directory

is generated.

It appears that the makefile is not generating instructions to build the
liboctave_fixed.dll.a, while the install is requiring it.  The library
liboctave_fixed.dll _is_  present in the main/fixed directory.  The only
workaround I have found is to create main/fixed/NOINSTALL, so the main/fixed
libraries will not be installed.  

Since the build seemed to be successful, it seems a shame not to install the
fixed point tools.  Can anyone point me to a workaround?

Thanks,

James R. Phillips

Re: [OctDev] getfields

[Andy A. <ad...@nc...>]

I say we can get rid of them. getfields is 'deprecated'
in the sense that we have Matlab compatible getfield/setfield
available. They are used in very few places in octave-forge,
and in non-speed critical ways.

Andy

----- Original Message -----
From: Teemu Ikonen <tpi...@pc...>
Date: Wednesday, September 29, 2004 3:29 am
Subject: [OctDev] getfields

> Hi all,
> 
> main/struct/getfields.oct dumps core with recent CVS versions of 
> octave and
> octave-forge. Since there are working m-file versions of getfields and
> setfields in the same directory, is there any reason to keep the
> oct-versions around?

[OctDev] getfields

[Teemu I. <tpi...@pc...>]

Hi all,

main/struct/getfields.oct dumps core with recent CVS versions of octave and
octave-forge. Since there are working m-file versions of getfields and
setfields in the same directory, is there any reason to keep the
oct-versions around?

Teemu

Re: [OctDev] Problem compiling octave-forge

[Paul K. <pki...@us...>]

2004-09-09 is for 2.1.57 and up.

You are welcome to provide backport patches to octave-forge.  I don't
mind supporting older versions, but have no interest in writing the
code to do it myself.

I've made it so that the compile will complete even if it is not correct
so assuming the particular function you need works with your version
of octave then you will be fine.

Paul

On Sep 23, 2004, at 3:14 PM, Ralph wrote:

>
> The file rand.cc in FIXES causes this error:
> --------------------------------------------------------
> g++ -c -fPIC -I/usr/include/octave-2.1.50 
> -I/usr/include/octave-2.1.50/octave
> -mieee-fp -O3 -march=athlon-xp -fomit-frame-pointer -pipe 
> -DHAVE_OCTAVE_21
> -DALLBITS rand.cc -o rand.o
> rand.cc: In function `void do_size(octave_value_list, int&, int&)':
> rand.cc:139: error: `xisnan' undeclared (first use this function)
> ---------------------------------------------------------
>
> I'm using Octave V2.1.50 and octave-forge from 2004-09-09

[OctDev] Re: Problem compiling octave-forge

[David B. <Dav...@mo...>]

xisnan is defined in lo-mappers.h, so it seems this isn't being included
correctly in 2.1.50 version of oct.h... Then I suggest adding

#include <octave/lo-mappers.h>

to rand.cc

D.



-- 
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

[OctDev] Problem compiling octave-forge

[Ralph <oct...@sp...>]

The file rand.cc in FIXES causes this error:
--------------------------------------------------------
g++ -c -fPIC -I/usr/include/octave-2.1.50 -I/usr/include/octave-2.1.50/octave 
-mieee-fp -O3 -march=athlon-xp -fomit-frame-pointer -pipe -DHAVE_OCTAVE_21 
-DALLBITS rand.cc -o rand.o
rand.cc: In function `void do_size(octave_value_list, int&, int&)':
rand.cc:139: error: `xisnan' undeclared (first use this function)
---------------------------------------------------------

I'm using Octave V2.1.50 and octave-forge from 2004-09-09

Ralph

-- 

Re: [OctDev] strcmp problem in interp1

[John W. E. <jw...@be...>]

On 23-Sep-2004, David Bateman <Dav...@mo...> wrote:

| Ok, I take it back... Its not a bug in octave-forge, but should be
| considered a bug in 2.1.59 as in matlab R12 I get
| 
| strcmp(NaN, 'test')
| 
| ans =
| 
|      0
| 
| >> strcmp(1,1)
| 
| ans =
| 
|      0
| 
| 
| So numeric arguments are acceptable to matlab and always return
| false...

How about the following change then?

jwe


scripts/ChangeLog:

2004-09-23  John W. Eaton  <jw...@oc...>

	* strings/strcmp.m: If args are not strings or cell arrays of
	strings, return zero instead of reporting an error.

 
Index: scripts/strings/strcmp.m
===================================================================
RCS file: /usr/local/cvsroot/octave/scripts/strings/strcmp.m,v
retrieving revision 1.24
diff -u -r1.24 strcmp.m
--- a/scripts/strings/strcmp.m	15 Sep 2004 18:28:56 -0000	1.24
+++ b/scripts/strings/strcmp.m	23 Sep 2004 17:25:20 -0000
@@ -139,11 +139,7 @@
       else
 	error ("strcmp: nonconformant cell arrays");
       endif
-    else
-      error ("strcmp: expecting args to be strings or cell arrays of strings");
     endif
-  else
-    error ("strcmp: expecting args to be strings or cell arrays of strings");
   endif
 
 endfunction

Re: [OctDev] strcmp problem in interp1

[David B. <Dav...@mo...>]

Ok, I take it back... Its not a bug in octave-forge, but should be
considered a bug in 2.1.59 as in matlab R12 I get

strcmp(NaN, 'test')

ans =3D

     0

>> strcmp(1,1)

ans =3D

     0


So numeric arguments are acceptable to matlab and always return
false...

D.

Dapr=E8s David Bateman <Dav...@mo...> (le 23/09/2004):
> There is a problem in interp1 where if the fifth argument is not
> defined it is given the value NaN. strcmp is then called as
>=20
> strcmp(NaN, "extrap")
>=20
> which as of 2.1.59 is not valid... Are there other places this occurs?
> Should I fix it or would you like to do it Paul? This is a rather nasty
> bug with the latest versions of octave and octave-forge, so does it imp=
ly
> a rapid release being needed?
>=20
> Cheers
> David
>=20
>=20
> --=20
> David Bateman                                Dav...@mo...
> Motorola CRM                                 +33 1 69 35 48 04 (Ph)=20
> Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax)=20
> 91193 Gif-Sur-Yvette FRANCE
>=20
> The information contained in this communication has been classified as:=
=20
>=20
> [x] General Business Information=20
> [ ] Motorola Internal Use Only=20
> [ ] Motorola Confidential Proprietary
>=20
>=20
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev

--=20
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph)=20
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax)=20
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as:=20

[x] General Business Information=20
[ ] Motorola Internal Use Only=20
[ ] Motorola Confidential Proprietary

[OctDev] strcmp problem in interp1

[David B. <Dav...@mo...>]

There is a problem in interp1 where if the fifth argument is not
defined it is given the value NaN. strcmp is then called as

strcmp(NaN, "extrap")

which as of 2.1.59 is not valid... Are there other places this occurs?
Should I fix it or would you like to do it Paul? This is a rather nasty
bug with the latest versions of octave and octave-forge, so does it imply
a rapid release being needed?

Cheers
David


-- 
David Bateman                                Dav...@mo...
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

Re: [OctDev] gnumericread, gnumericfinfo

[Josep i T. <jm...@pu...>]

On dj, 2004-09-23 at 00:32, Jonathan Stickel wrote:
> I have tried gnumeric, and it does seem like a satisfactory spreadsheet.
>   IIRC, it's graphing was better than OO.o.  However, I use OO.o when I
> need traditional office software (including spreadsheets) due to its
> multi-platform support.  Whatever I do in OO.o I can teach to someone
> using OO.o in Windows.

Yes... that's true... what I like about gnumeric is that it can handle
very large files. Larger than OOo or Excel can (in the same computer, of
course). It also has a very good support of Excel files. I don't expect
very sophisticated mathematical functions in a spreadsheet. For that I
use Octave.

>=20
> Your idea of an interface between spreadsheet software and octave is a
> good one, though.  Spreadsheets are great for visualizing data, but
> operating on that data can be difficult, especially when lists are
> variable lengths.  Hope it goes well for you!


hehe... I'll use it to process data that has been introduced by other
people (which find anything that sounds like programming to difficult,
or uninteresting, for them). Well... actually for my girlfriend's
thesis.

BTW, I didn't have any idea... This is just like xlsread (and behaviour
should be compatible with it), but for a better spreadsheet :)


Regards!
--=20
Josep Mon=E9s i Teixidor
Clau GnuPG: gpg --recv-keys 80E85CC4

Re: [OctDev] gnumericread, gnumericfinfo

[Jonathan S. <jjs...@sb...>]

I have tried gnumeric, and it does seem like a satisfactory spreadsheet.
  IIRC, it's graphing was better than OO.o.  However, I use OO.o when I
need traditional office software (including spreadsheets) due to its
multi-platform support.  Whatever I do in OO.o I can teach to someone
using OO.o in Windows.

Your idea of an interface between spreadsheet software and octave is a
good one, though.  Spreadsheets are great for visualizing data, but
operating on that data can be difficult, especially when lists are
variable lengths.  Hope it goes well for you!

Jonathan


Josep Monés i Teixidor wrote:
<snip>
> Does someone besides me use gnumeric here? :)

[OctDev] gnumericread, gnumericfinfo

[Josep i T. <jm...@pu...>]

Hi!

I have finish gnumericread and gnumericfinfo (well almost, there are a
lot of things that need to be polished, including docs) and I would like
to include them to gnumeric, but I have a few questions:

1.- They depend on libxml2, is this a problem?

2.- Where should they lie in the directory structure? main/io ? Or
perhaps extra/io ?

3.- A matter of taste: what do people prefer, gnumericread or gmrread?=20


It uses libxml2 xmlreader interface (a kind of sax interface but using a
push API instead of a pull API); so it should be fast and use very few
resources.

I plan to include soon a gnumericwrite function (but it's a whole
different because it must keep a tree of the whole document) and
probably some functions to octave specific things.

If someone wants to test them you can download them here:
http://www.somvius.net/gnumeric_octave.tar.gz

Does someone besides me use gnumeric here? :)
--=20
Josep Mon=E9s i Teixidor
Clau GnuPG: gpg --recv-keys 80E85CC4

[OctDev] Building Octave on Cygwin with dyanamically linked atlas libraries

[James R. P. <JRP...@co...>]

Hi,

I recently built octave 2.1.58 from source on cygwin quite successfully, 
using Paul Soderlin's method of linking to static atlas (.a) libraries, 
optimized for my processor.  I was wondering if anyone could provide 
guidance on how to build atlas as a dynamically loaded library (.so?) 
under Cygwin, and then link octave and octave-forge against that.  This 
would allow building only one octave-forge binary, and then just 
selecting the correct optimized version of atlas to install based on 
your cpu architecture.

James R. Phillips
South Bend, Indiana

Re: [OctDev] Problems with octaver-forge install

[Josep i T. <jm...@pu...>]

On dl, 2004-09-20 at 18:58, Tomasv wrote:
> Hi.
> This is a message from mi octave start-up
>=20
> _________________________________________________________________________=
__
> GNU Octave, version 2.1.57 (i686-pc-linux-gnu).
> Copyright (C) 2004 John W. Eaton.
> This is free software; see the source code for copying conditions.
> There is ABSOLUTELY NO WARRANTY; not even for MERCHANTIBILITY or
> FITNESS FOR A PARTICULAR PURPOSE.  For details, type `warranty'.
>=20
> Additional information about Octave is available at http://www.octave.org=
.
>=20
> Please contribute if you find this software useful.
> For more information, visit http://www.octave.org/help-wanted.html
>=20
> Report bugs to <bug...@be...> (but first, please read
> http://www.octave.org/bugs.html to learn how to write a helpful report).
>=20
> error:
> /usr/local/libexec/octave/2.1.57/site/oct/i686-pc-linux-gnu/octave-forge/=
dispatch.oct
> is not a valid shared library
> error: `dispatch' undefined near line 2 column 1
> error: near line 2 of file
> `/usr/local/share/octave/2.1.57/site/m/octave-forge/fixed//PKG_ADD'
> error: source: error sourcing file
> `/usr/local/share/octave/2.1.57/site/m/octave-forge/fixed//PKG_ADD'
> error: near line 7 of file `/home/evalero/.octaverc'
> octave:1>
> _________________________________________________________________________=
__



I had the same problem: I solved it by compiling both Octave and
Octave-Forge using the following flags to configure:

--enable-shared --disable-static

I don't know exactly what they do (does static compile all liboctave
statically to each oct?), but that worked...


Good luck,
--=20
Josep Mon=E9s i Teixidor
Clau GnuPG: gpg --recv-keys 80E85CC4

[OctDev] Problems with octaver-forge install

[Tomasv <to...@la...>]

Hi.
This is a message from mi octave start-up

___________________________________________________________________________
GNU Octave, version 2.1.57 (i686-pc-linux-gnu).
Copyright (C) 2004 John W. Eaton.
This is free software; see the source code for copying conditions.
There is ABSOLUTELY NO WARRANTY; not even for MERCHANTIBILITY or
FITNESS FOR A PARTICULAR PURPOSE.  For details, type `warranty'.

Additional information about Octave is available at http://www.octave.org.

Please contribute if you find this software useful.
For more information, visit http://www.octave.org/help-wanted.html

Report bugs to <bug...@be...> (but first, please read
http://www.octave.org/bugs.html to learn how to write a helpful report).

error:
/usr/local/libexec/octave/2.1.57/site/oct/i686-pc-linux-gnu/octave-forge/dispatch.oct
is not a valid shared library
error: `dispatch' undefined near line 2 column 1
error: near line 2 of file
`/usr/local/share/octave/2.1.57/site/m/octave-forge/fixed//PKG_ADD'
error: source: error sourcing file
`/usr/local/share/octave/2.1.57/site/m/octave-forge/fixed//PKG_ADD'
error: near line 7 of file `/home/evalero/.octaverc'
octave:1>
___________________________________________________________________________

This is mi .octaverc

___________________________________________________________________________
## System-wide startup file for Octave.
##
## This file should contain any commands that should be executed each
## time Octave starts for every user at this site.


LOADPATH = [
'/usr/local/libexec/octave/2.1.57/site/oct/i686-pc-linux-gnu/octave-forge:/usr/local/share/octave/2.1.57/site/m/octave-forge//:',
LOADPATH ];

EXEC_PATH = [
'/usr/local/libexec/octave/2.1.57/site/exec/i686-pc-linux-gnu:', EXEC_PATH
];
___________________________________________________________________________

Mi OS is Fedora Core 2

--
Tomás V. Orellana
to...@la...
Labbi/DCET/UESC