octave-dev — GNU Octave developers' list

[OctDev] changes in optim

[Michael C. <mic...@ua...>]

Hello all,
I have made some significant changes in the ./main/optim directory.

bfgs.m and bs_gradient.m have been replaced with bfgsmin.cc and 
numgradient.cc. The first change introduces interface incompatibilities, but 
the second has a common interface.

There will be some errors in functions that call bfgs.m until the new 
interface and function name are adopted.

There are also two new minimization routines: lbfgsmin.cc and samin.cc, that 
implement limited memory BFGS, and Simulated Annealing.

There is also a celleval.cc, which is like leval.cc, but for cells. Perhaps 
this might be useful in contexts other that optimization.

The motivation for the change is the greater speed and reliability of the new 
routines. I can post documentation if anyone is interested.

I'm sorry for any inconveniences adapting interfaces might cause. I have kept 
snapshots of the replaced functions if anyone needs them.

Regards, Michael

[OctDev] rgb2ycbcr yuv2rgb

[Dhoto <dh...@ao...>]

Any one has this function for octave

rgb2ycbcr yuv2rgb

where can i find those function

thank you

-- 
Best regard,

--------------------------------------------@
dh...@ao...

Re: [OctDev] someone please add these files

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

Oops, misplaced backquote in the find command below:

	for f in `find . -name Root -print`; do mv $f $f.in; sed 
-e's/[.]octave//' <$f.in > $f; rm -f $f.in; done

Paul Kienzle
pki...@us...

On May 14, 2004, at 5:53 AM, Paul Kienzle wrote:

> The error I was getting is the following:
>
> 	Cannot access /cvsroot/octave/CVSROOT
> 	No such file or directory
>
> I fetched a new copy of the entire archive.
>
> The file CVS/Root has changed:
>
> 	Avocado:~/cvs paulkienzle$ cat octave-forge-old/CVS/Root
> 	:ext:pki...@cv...:/cvsroot/octave
> 	Avocado:~/cvs paulkienzle$ cat octave-forge/CVS/Root
> 	:ext:pki...@cv...:/cvsroot/octave
>
> Removing .octave from the address if octave-forge-old seems to
> work, so I did the following:
>
> 	for f in `find . -name Root` -print; do mv $f $f.in; sed 
> -e's/[.]octave//' <$f.in > $f; rm -f $f.in; done
>
> Let me know if that works.
>
> Paul Kienzle
> pki...@us...
>
> On May 14, 2004, at 5:28 AM, Michael Creel wrote:
>
>> For some reason, octave-forge isn't letting me commit files. I can 
>> log into my
>> sourceforge account as usual, but can't commit to octave-forge. There 
>> is a
>> small bug in the two attached files that I'd like to fix. Would 
>> someone
>> please commit them to main/optim/mintoolkit
>>
>> Thanks, Michael
>> <lbfgsmin.cc><bfgsmin.cc>
>

[OctDev] NO, don't commit them

[Michael C. <mic...@ua...>]

I'm sorry to bother everyone. The attached files seem to have word-wrapped 
somewhere in transmission, and they won't compile. Forget it, I'll try to 
figure out why I can't commit.  Michael

Re: [OctDev] someone please add these files

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

The error I was getting is the following:

	Cannot access /cvsroot/octave/CVSROOT
	No such file or directory

I fetched a new copy of the entire archive.

The file CVS/Root has changed:

	Avocado:~/cvs paulkienzle$ cat octave-forge-old/CVS/Root
	:ext:pki...@cv...:/cvsroot/octave
	Avocado:~/cvs paulkienzle$ cat octave-forge/CVS/Root
	:ext:pki...@cv...:/cvsroot/octave

Removing .octave from the address if octave-forge-old seems to
work, so I did the following:

	for f in `find . -name Root` -print; do mv $f $f.in; sed 
-e's/[.]octave//' <$f.in > $f; rm -f $f.in; done

Let me know if that works.

Paul Kienzle
pki...@us...

On May 14, 2004, at 5:28 AM, Michael Creel wrote:

> For some reason, octave-forge isn't letting me commit files. I can log 
> into my
> sourceforge account as usual, but can't commit to octave-forge. There 
> is a
> small bug in the two attached files that I'd like to fix. Would someone
> please commit them to main/optim/mintoolkit
>
> Thanks, Michael
> <lbfgsmin.cc><bfgsmin.cc>

[OctDev] someone please add these files

[Michael C. <mic...@ua...>]

For some reason, octave-forge isn't letting me commit files. I can log into my 
sourceforge account as usual, but can't commit to octave-forge. There is a 
small bug in the two attached files that I'd like to fix. Would someone 
please commit them to main/optim/mintoolkit

Thanks, Michael

[OctDev] pwelch.m bug report and patch

[<rol...@ie...>]

                                Hello,
I hope this is the right place to report bugs in octave-forge.

pwelch.m does not handle complex row vectors correctly.  This is
demonstrated in the script below.  White noise is filtered by a filter
with complex impulse response.  The filter frequency response peaks at
f/fs = 0.2 and so does the spectrum of the output signal.  However, if
the signal is a row vector, pwelch and tfe yield a peak at f/fs = 0.8.
The patch below (for pwelch.m as of octave-forge 2001.11.02, as found
in Debian octave-forge   2001.11.02-4) corrects this.  I believe that
the problem exists also in the code as of a few days ago, but I cannot
easily test it because I run rather old versions of Octave.
                                        hope this helps
                                               Roland

P.S. I am not (yet) subscribed to octave-dev, please CC me. Thanks.

----------- script to demonstrate the problem -----------
## pwelch_bug.m
## Created: 2004-05-11T10:48:25+0200  rol...@ie...
##   This code is put in the public domain.
## Demonstrate bug (incorrect handling of complex row vectors) in
## pwelch.m

Nresp = 60;                     # impulse response length
Wresp = 0.1;                    # filter bandwidth
h = exp(-Wresp * (0:(Nresp-1)));
Nfft = 256;
[H, w] = freqz(h, 1, Nfft);     # Frequency response

## Shift the filter response to F0.
F0 = 0.2;                       # centre frequency
hs = h .* exp(2i * pi * F0 * (0:(Nresp-1)));
Hs = freqz(hs, 1, Nfft);        # Frequency response

multiplot(2, 2);

## Generate input signal
Nsig = 10000;                   # signal length
x = 0.5 * randn(1, Nsig);
y = filter(hs, 1, x);

[Syy, Faxis]  = pwelch(y, Nfft, 1, 'whole'); # PSD of y
[Syyc, Faxis] = pwelch(y .', Nfft, 1, 'whole'); # PSD of y

Sxy  = tfe(x, y, Nfft, 1, 'whole');
Sxyc = tfe(x, y .', Nfft, 1, 'whole');

subplot(221);
clearplot;
title('Filter Magnitude');
xlabel('f/fs');
ylabel('20 log|H(f)| [dB]');
plot(w/pi, 20*log10(abs(Hs)), ';Filter response;');
hold on
plot(Faxis, 10*log10(abs(Syyc)), ';Syy col;');
plot(Faxis, 20*log10(abs(Sxyc)), '+;Sxy col;');

#subplot(222);
#clearplot;
#title('Filter Magnitude, row');
#xlabel('f/fs');
#ylabel('20 log|H(f)| [dB]');
#plot(w/pi, 20*log10(abs(Hs)), ';Filter response;');
plot(Faxis, 10*log10(abs(Syy)), ';Syy row;');
plot(Faxis, 20*log10(abs(Sxy)), '+;Sxy row;');

## Check filter phase
subplot(222);
clearplot;
title('Filter Phase, column');
xlabel('f/fs');
ylabel('arg(H(f)) [rad]');
plot(Faxis, unwrap(arg(Hs)), ';Filter phase;');

## Note that the cross-spectral density Sxy and the
## filter frequency response H are related as
## Sxy(f) = conj(H(f)):
## 
## Sxy(f) = Sxx(f) H^*(f)
## Syy(f) = Sxy(f) H(f) = Sxx(f) |H(f)|^2
## 
## Athanasios Papoulis
## Signal Analysis
## isbn 0-07-048460-0
## McGraw-Hill, 1984
## 9.3 Spectral Analysis, p. 312

plot(Faxis, -unwrap(arg(Sxyc)), '+;tfe phase;');
## plot(Faxis, -unwrap(arg(Sxy)), ';tfe phase row;');

subplot(223);
clearplot;
title('Estimate error, column');
xlabel('f/fs');
ylabel('20 log|H_est(f)/H(f)| [dB]');
plot(w/pi, 20*log10(abs(Sxyc .')) - 20*log10(abs(Hs)), ';Magnitude error;');

subplot(224);
clearplot;
title('Estimate error, column');
xlabel('f/fs');
ylabel('arg(H_est(f)/H(f)) [rad]');
plot(w/pi, -unwrap(arg(Sxyc .'))-unwrap(arg(Hs)), ';Phase error;');

## pwelch_bug.m ends here.


--------- patch against pwelch.m as of octave-forge 2001.11.02 ---------
*** pwelch.m	Wed Jan 23 03:05:25 2002
--- pwelch.m	Thu May 13 11:44:08 2004
***************
*** 99,104 ****
      error ([calledby, " data must be a vector"]);
    end
!   if columns(x) != 1, x = x'; end
!   if columns(y) != 1, y = y'; end
    if !isempty(y) && rows(x)!=rows(y)
      error ([calledby, " x and y vectors must be the same length"]);
--- 99,105 ----
      error ([calledby, " data must be a vector"]);
    end
!   ## Use simple transposition to avoid conjugation of complex x, y
!   if columns(x) != 1, x = x .'; end
!   if columns(y) != 1, y = y .'; end
    if !isempty(y) && rows(x)!=rows(y)
      error ([calledby, " x and y vectors must be the same length"]);

----------------- patch ends above this line ----------------------

[OctDev] Skipping over text headers with dlmread

[Mark P. E. <ma...@sr...>]

	I had a need to import into octave text files with with headers.  I modified 
dlmread.m so that it could skip over lines of text before reading the data.  
The addition should be compatible with Matlab even though it doesn't 
implement all the functionality.  I don't have Matlab so I couldn't check for 
sure.  I also fixed a bug in dlmread that made it so some deliminators like 
";" wouldn't work.  I have attached a patch.  Feel free to modify it as 
needed, I am not very experienced at programing m-files.

                                    -Mark Esplin

--- /usr/local/share/octave/2.1.57/site/m/octave-forge/io/dlmread.m     
2004-05-07 15:00:21.000000000 -
0400
+++ dlmread.m   2004-05-07 14:59:39.000000000 -0400
@@ -14,9 +14,10 @@
 ## along with this program; if not, write to the Free Software
 ## Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA

-## x = dlmread (filename, sep)
+## x = dlmread (filename, sep, row)
 ##    Read the matrix x from a file, with columns separated by the
 ##    character sep (default is ",").  NaN values are written as nan.
+##    The number of rows to skip before reading data is row (default is 0).
 ##
 ## WARNING: for compatibility, must treat empty fields as zero, but doesn't.

@@ -24,12 +25,13 @@
 ## 2001-02-16
 ##    * first revision
-function x = dlmread (filename, sep)
+function x = dlmread (filename, sep, row)

-  if (nargin < 1 || nargin > 2)
-    usage ("x = dlmread (filename, sep)");
+  if (nargin < 1 || nargin > 3)
+    usage ("x = dlmread (filename, sep, row)");
   endif
-  if nargin < 3, sep = ","; endif
+  if nargin < 2, sep = ","; endif
+  if nargin < 3, row = 0; endif

   fid = fopen(filename, "r");
   if (fid >= 0)
@@ -45,18 +47,28 @@
     if (nr > 0)
        in(idx) = " ";
        nr += (idx(length(idx)) < length(in));
+        idxl = idx;
     endif
     idx = find (in == "\r");
     if (nr == 0)
        nr = length(idx);
        if (nr > 0) nr += (idx(length(idx)) < length(in)); endif
+       idxl = idx;
     endif
     if (length (idx) > 0) in(idx) = " "; endif
+    nr = nr-row;

+    ## find where to start reading data
+    if (row > 0)
+       istr = idxl(row)+1;
+    else
+       istr = 1;
+    endif
+
     ## convert separators to spaces
     idx = find (in == sep);
     if (length(idx) > 0) in(idx) = " "; endif
-    [x, n, err] = sscanf(in, "%g");
+    [x, n, err] = sscanf(in(istr:length(in)), "%g");
     if (!isempty(err))
       error(["dlmread: ", err]);
     elseif (rem(n, nr) != 0)

Re: [OctDev] result from function problem

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

The problem was reported already in the thread.

http://www.octave.org/mailing-lists/bug-octave/2004/294
http://www.octave.org/mailing-lists/bug-octave/2004/295
http://www.octave.org/mailing-lists/bug-octave/2004/298
http://www.octave.org/mailing-lists/bug-octave/2004/302
http://www.octave.org/mailing-lists/bug-octave/2004/307
http://www.octave.org/mailing-lists/bug-octave/2004/308

The conclusion, is that the problem is known, but that there is no
easy solution... Of course you're welcome to send a patch :-)

Sorry I don't know an easy workaround, though there are lots of
potential ugly workarounds to check upto a certain number of 
possible output args..

Cheers
David

According to BrI <oc...@th...> (on 05/05/04):
> 
> Hi,
> 
> I'm encountering a problem in Octave, and I'm wonder whether it's a bug, or
> rather just an interpretation of how some code should work.
> 
> I have a function:
> 
> function [a,b,c] = func1
> a = 100;
> b = 'fred';
> c = 300;
> 
> -------------
> 
> In Matlab:
> 
> [t{1:3}]=func1
> t =    [100]    'fred'    [300]
> 
> In Octave:
> 
> [t{1:3}]=func1
> t  =
> {
>    [1,1] [100]
>    [1,2] [100]
>    [1,3] [100]
> }
> 
> 
> So MatLab intreprets the range to mean that it should return more parameters
> into the cell, whereas Octave treats it to mean that the first parameter
> should be returned into t 3 times.
> 
> The reason I'm using this is because I'm using the function eval and I don't
> know how many parameters the function being evaluated will be returning, so
> I want to check them.  Does any body know of a good workaround in Octave?
> 
> cheers.
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: Oracle 10g
> Get certified on the hottest thing ever to hit the market... Oracle 10g. 
> Take an Oracle 10g class now, and we'll give you the exam FREE.
> http://ads.osdn.com/?ad_id=3149&alloc_id=8166&op=click
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev

-- 
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] result from function problem

[BrI <oc...@th...>]

Hi,

I'm encountering a problem in Octave, and I'm wonder whether it's a bug, or
rather just an interpretation of how some code should work.

I have a function:

function [a,b,c] = func1
a = 100;
b = 'fred';
c = 300;

-------------

In Matlab:

[t{1:3}]=func1
t =    [100]    'fred'    [300]

In Octave:

[t{1:3}]=func1
t  =
{
   [1,1] [100]
   [1,2] [100]
   [1,3] [100]
}


So MatLab intreprets the range to mean that it should return more parameters
into the cell, whereas Octave treats it to mean that the first parameter
should be returned into t 3 times.

The reason I'm using this is because I'm using the function eval and I don't
know how many parameters the function being evaluated will be returning, so
I want to check them.  Does any body know of a good workaround in Octave?

cheers.

[OctDev] GSL special functions

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

Hi all,

I've made Octave bindings to a large part of the special functions in the
GNU Scientific library (http://www.gnu.org/software/gsl/). The code can be
found from the Octave-forge CVS (http://octave.sf.net/) under main/gsl.

The bindings are made semi-automatically from a few template files, which
causes the docstrings to be a little too generic sometimes (i.e. the integer
indices of spherical harmonics are named n,m instead of traditional l,m),
otherwise the calling convention is Octave-like: Two output arguments result
in the error estimate being returned in the second value and input arguments
of differing sizes in the multi-argument functions are treated somewhat
intelligently.

The binding code is mostly trivial, but there's still a possibility of bugs
present, perhaps even in the underlying GSL code. Therefore, I'd eventually
like to have some tests to check for correct values in a few well selected
points in the functions. Does anyone know of an accurate source of such
values for a good number of special functions?

Teemu

Re: [OctDev] tk_octave problems

[Mark P. E. <ma...@sr...>]

Paul,

With the unsafe code and other problems it doesn't sound like tk_octave is the 
way to go.  I also tried soctcl which worked O.K. for  computations as long 
as plotting and file operations were performed from the tcl part of the code.  
Using soctcl, the octave ploting routines don't work, also pngread didn't 
work. I didn't try any other octave IO commands, but assumed that all IO will 
need to be provided by the tcl/tk part of the code. I guess I was looking 
more for a GUI for octave instead of a compute engine for tcl/tk.  After 
rereading your response I think I am starting to realize a little of the 
difficulty of putting a GUI on octave.  For now I will work with soctcl some 
more.

                       Thanks for the help,

                          -Mark Esplin

On Friday 23 April 2004 22:53, Paul Kienzle wrote:
> Mark,
>
> tk_octave is mostly unsupported.
>
> The biggest issue was that it required a version of octave
> compiled with pthreads, which would make deployment of
> my package more troublesome.   Instead I investigated
> other solutions.   For now I'm using sockets with listen and
> extra/soctcl from octave-forge.  This is adequate for my needs,
> but is considerably slower than it could be.
>
> If you decide to pursue tk_octave, please bear in mind that
> the code is not safe as it is currently written.   Since octave's
> reference counting classes do not use semaphores, there
> is a small window in which a task switch could delete
> an array in octave while tcl wanted to access it.  The simplest
> solution to this problem would be to use a separate namespace
> for shared variables which have proper threaded access
> semantics.
>
> You may want to consider unifying soctcl and tk_octave syntax.
> That way you can use network transparency if needed, but
> use threads and shared memory for speed.
>
> The soctcl interface is not as good as it could be.   I find
> that from tcl, I want some variables in octave to be scoped
> according to the current function scope in Tcl (and possibly
> persistent), and others to be global.  Some sort of namespace
> parameter for the octave eval command would do wonders.
> It would be easy enough to implement in octave by associating
> a symbol table with each namespace.
>
> An alternative to tk_octave is to embed octave in tcl.  A lot of
> the mechanics of this are already available.  Fundamentally
> this is no different from embedding tcl in octave, but in practice
> it is easier to extend tcl/tk with numeric support than it will be
> to extend octave with gui support.  As a bonus, making octave
> embeddable in tcl will remove the cygwin dependency in
> windows.
>
> Paul Kienzle
> pki...@us...
>
> On Apr 22, 2004, at 11:45 AM, Mark P. Esplin wrote:
> > I tried out tk_octave and noticed a couple of problems.  First on my
> > system
> > the tk_matrix demo wouldn't work until I added:
> >
> > set auto_path "$auto_path /usr/lib"
> >
> > to tk_matrix.tcl so that it could find the BLT package.  "wish"
> > doesn't have
> > any trouble finding the BLT package.
> >
> > When I start the tcl/tk interpreter from octave with tk_interp my CPU
> > free
> > time drops to 0.  This is before I do any other commands. Why does the
> > interpreter require so much CPU time?
> >
> > tk_entry.m doesn't work.  Line 100 is "@@ -98,6 +98,7 @@" which looks
> > like it
> > must of gotten left over from a patch attempt.
> >
> > The document string on tk_end says "See also: tk_start, ..." which
> > must have
> > supposed to have been tk_interp.
> >
> >                          -Mark Esplin
> >
> >
> >
> > -------------------------------------------------------
> > This SF.Net email is sponsored by: IBM Linux Tutorials
> > Free Linux tutorial presented by Daniel Robbins, President and CEO of
> > GenToo technologies. Learn everything from fundamentals to system
> > administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> > _______________________________________________
> > Octave-dev mailing list
> > Oct...@li...
> > https://lists.sourceforge.net/lists/listinfo/octave-dev
>
> -------------------------------------------------------
> This SF.net email is sponsored by: The Robotic Monkeys at ThinkGeek
> For a limited time only, get FREE Ground shipping on all orders of $35
> or more. Hurry up and shop folks, this offer expires April 30th!
> http://www.thinkgeek.com/freeshipping/?cpg=12297
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev

Re: [OctDev] tk_octave problems

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

Mark,

tk_octave is mostly unsupported.

The biggest issue was that it required a version of octave
compiled with pthreads, which would make deployment of
my package more troublesome.   Instead I investigated
other solutions.   For now I'm using sockets with listen and
extra/soctcl from octave-forge.  This is adequate for my needs,
but is considerably slower than it could be.

If you decide to pursue tk_octave, please bear in mind that
the code is not safe as it is currently written.   Since octave's
reference counting classes do not use semaphores, there
is a small window in which a task switch could delete
an array in octave while tcl wanted to access it.  The simplest
solution to this problem would be to use a separate namespace
for shared variables which have proper threaded access
semantics.

You may want to consider unifying soctcl and tk_octave syntax.
That way you can use network transparency if needed, but
use threads and shared memory for speed.

The soctcl interface is not as good as it could be.   I find
that from tcl, I want some variables in octave to be scoped
according to the current function scope in Tcl (and possibly
persistent), and others to be global.  Some sort of namespace
parameter for the octave eval command would do wonders.
It would be easy enough to implement in octave by associating
a symbol table with each namespace.

An alternative to tk_octave is to embed octave in tcl.  A lot of
the mechanics of this are already available.  Fundamentally
this is no different from embedding tcl in octave, but in practice
it is easier to extend tcl/tk with numeric support than it will be
to extend octave with gui support.  As a bonus, making octave
embeddable in tcl will remove the cygwin dependency in
windows.

Paul Kienzle
pki...@us...

On Apr 22, 2004, at 11:45 AM, Mark P. Esplin wrote:

> I tried out tk_octave and noticed a couple of problems.  First on my 
> system
> the tk_matrix demo wouldn't work until I added:
>
> set auto_path "$auto_path /usr/lib"
>
> to tk_matrix.tcl so that it could find the BLT package.  "wish" 
> doesn't have
> any trouble finding the BLT package.
>
> When I start the tcl/tk interpreter from octave with tk_interp my CPU 
> free
> time drops to 0.  This is before I do any other commands. Why does the
> interpreter require so much CPU time?
>
> tk_entry.m doesn't work.  Line 100 is "@@ -98,6 +98,7 @@" which looks 
> like it
> must of gotten left over from a patch attempt.
>
> The document string on tk_end says "See also: tk_start, ..." which 
> must have
> supposed to have been tk_interp.
>
>                          -Mark Esplin
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by: IBM Linux Tutorials
> Free Linux tutorial presented by Daniel Robbins, President and CEO of
> GenToo technologies. Learn everything from fundamentals to system
> administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev
>

[OctDev] tk_octave problems

[Mark P. E. <ma...@sr...>]

I tried out tk_octave and noticed a couple of problems.  First on my system 
the tk_matrix demo wouldn't work until I added:

set auto_path "$auto_path /usr/lib"

to tk_matrix.tcl so that it could find the BLT package.  "wish" doesn't have 
any trouble finding the BLT package.  

When I start the tcl/tk interpreter from octave with tk_interp my CPU free 
time drops to 0.  This is before I do any other commands. Why does the 
interpreter require so much CPU time?

tk_entry.m doesn't work.  Line 100 is "@@ -98,6 +98,7 @@" which looks like it 
must of gotten left over from a patch attempt.

The document string on tk_end says "See also: tk_start, ..." which must have 
supposed to have been tk_interp.

                         -Mark Esplin

Re: [OctDev] trouble compiling octave-forge

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

Yeah, its a known problem due to a change in versions of octave that
occured after the 2004.02.12 release. It is fixed in the CVS.

Regards
David

According to Mark P. Esplin <ma...@sr...> (on 04/21/04):
> 	I am having trouble compiling octave-forge-2004.02.12 on Fedora Core 1 linux. 
> It dies with the following errors:
> 
> galois.cc: In function `boolMatrix mx_el_lt(const galois&, const galois&)':
> galois.cc:1282: error: `NBM' undeclared (first use this function)
> galois.cc:1282: error: (Each undeclared identifier is reported only once for
>    each function it appears in.)
> galois.cc: In function `boolMatrix mx_el_eq(const galois&, const galois&)':
> galois.cc:1282: error: `TBM' undeclared (first use this function)
> galois.cc:1282: error: `FBM' undeclared (first use this function)
> make[2]: *** [galois.o] Error 1
> 	
> I wanted to try out tk_octave from source forge.  I compiled and installed 
> octave2.1.57 with -lpthread and that seemed to work. I deleted the NOINSTALL 
> file in extra/tk_octave and set the make file for my system (no VTK and I 
> have tk8.3 instead of tk8.4). That all seemed to work, but the compile died 
> further along.
> 
>                                          -Mark Esplin
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: IBM Linux Tutorials
> Free Linux tutorial presented by Daniel Robbins, President and CEO of
> GenToo technologies. Learn everything from fundamentals to system
> administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev

-- 
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] trouble compiling octave-forge

[Mark P. E. <ma...@sr...>]

	I am having trouble compiling octave-forge-2004.02.12 on Fedora Core 1 linux. 
It dies with the following errors:

galois.cc: In function `boolMatrix mx_el_lt(const galois&, const galois&)':
galois.cc:1282: error: `NBM' undeclared (first use this function)
galois.cc:1282: error: (Each undeclared identifier is reported only once for
   each function it appears in.)
galois.cc: In function `boolMatrix mx_el_eq(const galois&, const galois&)':
galois.cc:1282: error: `TBM' undeclared (first use this function)
galois.cc:1282: error: `FBM' undeclared (first use this function)
make[2]: *** [galois.o] Error 1
	
I wanted to try out tk_octave from source forge.  I compiled and installed 
octave2.1.57 with -lpthread and that seemed to work. I deleted the NOINSTALL 
file in extra/tk_octave and set the make file for my system (no VTK and I 
have tk8.3 instead of tk8.4). That all seemed to work, but the compile died 
further along.

                                         -Mark Esplin

[OctDev] Re: sparse: size and indexing

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

On Sat, 27 Mar 2004, Schloegl Alois wrote:

> I'm using the latest CVS-versions of Octave and Octave-forge and observed
> two problems using SPARSE matrices.
>
> (1) size of a sparse matrix can not be obtained, [-1,-1] is returned.

I can't repeat this (with 2.1.57 and yesterday's octave-forge)
octave:1> which sparse
sparse is the dynamically-linked function from the file
/usr/lib/octave/2.1.57/site/oct/i386-pc-linux-gnu/octave-forge/sparse.oct
octave:2> x=sparse(2:31,1:30,1,31,30);
octave:3> size(x)
ans =
  31  30
octave:4> H.x=sparse(2:31,1:30,1,31,30);
octave:5> size(H.x)
ans =
  31  30

> (2) indexing of a sparse field within a struct stops with an error.

I don't believe that this has ever worked.

octave:6> H.x(2:4,:)
error: can't perform indexing operations for sparse type
octave:6> x(2:4,:)
ans = ...


I would like to try to fix (2) if possible. Could someone
explain to me if dynamically loaded types need to do anything
special to allow indexing in a struct?

andy

[OctDev] documentation for mintoolkit minimization functions

[Michael C. <mic...@ua...>]

MINTOOLKIT is in the main/optim/mintoolkit directory of octave-forge. I
have written up some documentation, at
http://pareto.uab.es/mcreel/MINTOOLKIT/mintoolkit.pdf
Any comments are welcome. Michael

Re: [OctDev] help with CVS

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

Dapr=E8s Michael Creel <mic...@ua...> (le 31/03/2004):
> If anyone knows how to, could you please remove the directory=20
> main/optim/Optimize/ from octave-forge? I can't see how to do it myself=
.=20
> The .m files there are not going to be maintained, and their presence w=
ill=20
> just add to the problem of knowing which of the options in optim is lik=
ely to=20
> be the best for a given problem. The equivalents are in=20
> main/optim/mintoolkit, which I fully expect to maintain with stable=20
> interfaces.
> Thanks, Michael

With CVS unless you have access to the cvsroot tree itself there is no wa=
y to
remove a directiory. All you can do is empty the directory by removing al=
l of
the files. This is one of the many problems with CVS....

D.

--=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] help with CVS

[Michael C. <mic...@ua...>]

If anyone knows how to, could you please remove the directory 
main/optim/Optimize/ from octave-forge? I can't see how to do it myself. 
The .m files there are not going to be maintained, and their presence will 
just add to the problem of knowing which of the options in optim is likely to 
be the best for a given problem. The equivalents are in 
main/optim/mintoolkit, which I fully expect to maintain with stable 
interfaces.
Thanks, Michael

Re: [OctDev] Re: struct ("a",{1},"b",{1,2})

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

  Hi all,

  maybe we could have an ostruct() with the old behavior? If stinckin'
matlab doesn't do structs that take lists as values, should we be
deprived of them? For now I rm'd struct.cc and use good'ol struct.m
that behaves as I want.

  Etienne

On Tue, Mar 30, 2004 at 07:36:03PM -0500, Paul Kienzle wrote:
# Yes, I did do a number of compatibility changes.   struct with these
# changes is now part of octave.
# 
# Sorry for breaking things...
# 
# Does the new interface do everything you need?  Or is there
# some way we can extend it to do what you need?
# 
# I don't remember why I didn't enable the obvious behaviour of
# struct() creating an empty struct.
# 
# - Paul
# 
# On Mar 30, 2004, at 3:26 PM, Etienne Grossmann wrote:
# 
# >
# >  Hi again, sorry to answer to myself. I guess the effect I described
# >in my previous mail are due to an attempt to imitate the other brand,
# >which behaves like:
# >======================================================================
# >>>a = struct('x',{1},'y',{1,2})
# >
# >a =
# >1x2 struct array with fields:
# >    x
# >    y
# >
# >>>a.x
# >
# >ans =
# >     1
# >ans =
# >     1
# >
# >>>a.y
# >
# >ans =
# >     1
# >ans =
# >     2
# >======================================================================
# >
# >  Hmm. So no more structs taking lists as values? I mean 'cells as
# >values', since lists seem to be doomed?
# >
# >  Etienne
# >
# >On Tue, Mar 30, 2004 at 02:51:12PM -0500, Etienne Grossmann wrote:
# >#
# >#   Hi all,
# >#
# ># I use, octave 2.1.53, cvs octave-forge, so my prob could be due to 
# >not
# ># using 2.1.57.
# >#
# >#   Is that the desired behavior of struct()? Are those bugs introduced
# ># between r 1.2 and r 1.3 of struct.cc?
# >#
# ># 
# >======================================================================
# >#   octave:3> struct ("a",{1},"b",{1,2})
# >#   ans =
# >#   {  a =
# >#       (,
# >#         [1] = 1
# >#         [2] = 1
# >#       ,)
# >#
# >#     b =
# >#       (,
# >#         [1] = 1
# >#         [2] = 2
# >#       ,)
# >#   }
# >#
# >#   octave:4> struct ("a",{1},"b",{1,2},"c",{})
# >#   error: struct: dimensions of parameter 4 do not match those of 
# >parameter 6
# >#   octave:4> struct ()
# >#
# >#   *** struct:
# >#
# >#   struct('field',value,'field',value,...)
# >#
# >#   Create a structure and initialize its value.
# >#
# >#   struct('field',{values},'field',{values},...)
# >#
# >#   Create a structure array and initialize its values.  The dimensions
# >#   of each array of values must match.  Singleton cells and non-cell 
# >values
# >#   are repeated so that they fill the entire array.
# >#
# >#   struct('field',{},'field',{},...)
# >#
# ># Create an empty structure array.
# ># 
# >======================================================================
# >#
# >#   Once upon a time, struct() returned an empty struct, without need 
# >to
# ># do struct([]). No field-value pair -> empty struct, right?
# >#
# >#   I don't recall that there's anything wrong with having struct 
# >values
# ># being lists, my code has relied on this for some time now.
# >#
# >#   Any tips welcome,
# >#
# >#   Etienne
# >#
# ># --
# ># Etienne Grossmann ------ http://www.cs.uky.edu/~etienne
# >
# >-- 
# >Etienne Grossmann ------ http://www.cs.uky.edu/~etienne
# >
# >
# >-------------------------------------------------------
# >This SF.Net email is sponsored by: IBM Linux Tutorials
# >Free Linux tutorial presented by Daniel Robbins, President and CEO of
# >GenToo technologies. Learn everything from fundamentals to system
# >administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
# >_______________________________________________
# >Octave-dev mailing list
# >Oct...@li...
# >https://lists.sourceforge.net/lists/listinfo/octave-dev
# >
# 

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

Re: [OctDev] Re: struct ("a",{1},"b",{1,2})

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

Yes, I did do a number of compatibility changes.   struct with these
changes is now part of octave.

Sorry for breaking things...

Does the new interface do everything you need?  Or is there
some way we can extend it to do what you need?

I don't remember why I didn't enable the obvious behaviour of
struct() creating an empty struct.

- Paul

On Mar 30, 2004, at 3:26 PM, Etienne Grossmann wrote:

>
>   Hi again, sorry to answer to myself. I guess the effect I described
> in my previous mail are due to an attempt to imitate the other brand,
> which behaves like:
> ======================================================================
>>> a = struct('x',{1},'y',{1,2})
>
> a =
> 1x2 struct array with fields:
>     x
>     y
>
>>> a.x
>
> ans =
>      1
> ans =
>      1
>
>>> a.y
>
> ans =
>      1
> ans =
>      2
> ======================================================================
>
>   Hmm. So no more structs taking lists as values? I mean 'cells as
> values', since lists seem to be doomed?
>
>   Etienne
>
> On Tue, Mar 30, 2004 at 02:51:12PM -0500, Etienne Grossmann wrote:
> #
> #   Hi all,
> #
> # I use, octave 2.1.53, cvs octave-forge, so my prob could be due to 
> not
> # using 2.1.57.
> #
> #   Is that the desired behavior of struct()? Are those bugs introduced
> # between r 1.2 and r 1.3 of struct.cc?
> #
> # 
> ======================================================================
> #   octave:3> struct ("a",{1},"b",{1,2})
> #   ans =
> #   {  a =
> #       (,
> #         [1] = 1
> #         [2] = 1
> #       ,)
> #
> #     b =
> #       (,
> #         [1] = 1
> #         [2] = 2
> #       ,)
> #   }
> #
> #   octave:4> struct ("a",{1},"b",{1,2},"c",{})
> #   error: struct: dimensions of parameter 4 do not match those of 
> parameter 6
> #   octave:4> struct ()
> #
> #   *** struct:
> #
> #   struct('field',value,'field',value,...)
> #
> #   Create a structure and initialize its value.
> #
> #   struct('field',{values},'field',{values},...)
> #
> #   Create a structure array and initialize its values.  The dimensions
> #   of each array of values must match.  Singleton cells and non-cell 
> values
> #   are repeated so that they fill the entire array.
> #
> #   struct('field',{},'field',{},...)
> #
> # Create an empty structure array.
> # 
> ======================================================================
> #
> #   Once upon a time, struct() returned an empty struct, without need 
> to
> # do struct([]). No field-value pair -> empty struct, right?
> #
> #   I don't recall that there's anything wrong with having struct 
> values
> # being lists, my code has relied on this for some time now.
> #
> #   Any tips welcome,
> #
> #   Etienne
> #
> # --
> # Etienne Grossmann ------ http://www.cs.uky.edu/~etienne
>
> -- 
> Etienne Grossmann ------ http://www.cs.uky.edu/~etienne
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by: IBM Linux Tutorials
> Free Linux tutorial presented by Daniel Robbins, President and CEO of
> GenToo technologies. Learn everything from fundamentals to system
> administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> _______________________________________________
> Octave-dev mailing list
> Oct...@li...
> https://lists.sourceforge.net/lists/listinfo/octave-dev
>

[OctDev] Re: struct ("a",{1},"b",{1,2})

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

  Hi again, sorry to answer to myself. I guess the effect I described
in my previous mail are due to an attempt to imitate the other brand,
which behaves like:
======================================================================
>> a = struct('x',{1},'y',{1,2})

a = 
1x2 struct array with fields:
    x
    y

>> a.x

ans =
     1
ans =
     1

>> a.y

ans =
     1
ans =
     2
======================================================================

  Hmm. So no more structs taking lists as values? I mean 'cells as
values', since lists seem to be doomed?

  Etienne

On Tue, Mar 30, 2004 at 02:51:12PM -0500, Etienne Grossmann wrote:
# 
#   Hi all,
# 
# I use, octave 2.1.53, cvs octave-forge, so my prob could be due to not
# using 2.1.57.
# 
#   Is that the desired behavior of struct()? Are those bugs introduced
# between r 1.2 and r 1.3 of struct.cc?
# 
# ======================================================================
#   octave:3> struct ("a",{1},"b",{1,2})
#   ans =
#   {  a =
#       (,
#         [1] = 1
#         [2] = 1
#       ,)
# 
#     b =
#       (,
#         [1] = 1
#         [2] = 2
#       ,)
#   }
# 
#   octave:4> struct ("a",{1},"b",{1,2},"c",{})
#   error: struct: dimensions of parameter 4 do not match those of parameter 6
#   octave:4> struct ()
# 
#   *** struct:
# 
#   struct('field',value,'field',value,...)
# 
#   Create a structure and initialize its value.
# 
#   struct('field',{values},'field',{values},...)
# 
#   Create a structure array and initialize its values.  The dimensions
#   of each array of values must match.  Singleton cells and non-cell values
#   are repeated so that they fill the entire array.
# 
#   struct('field',{},'field',{},...)
# 
# Create an empty structure array.
# ======================================================================
# 
#   Once upon a time, struct() returned an empty struct, without need to
# do struct([]). No field-value pair -> empty struct, right?
# 
#   I don't recall that there's anything wrong with having struct values
# being lists, my code has relied on this for some time now.
# 
#   Any tips welcome,
# 
#   Etienne
# 
# -- 
# Etienne Grossmann ------ http://www.cs.uky.edu/~etienne

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

[OctDev] struct ("a",{1},"b",{1,2})

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

  Hi all,

I use, octave 2.1.53, cvs octave-forge, so my prob could be due to not
using 2.1.57.

  Is that the desired behavior of struct()? Are those bugs introduced
between r 1.2 and r 1.3 of struct.cc?

======================================================================
  octave:3> struct ("a",{1},"b",{1,2})
  ans =
  {  a =
      (,
        [1] = 1
        [2] = 1
      ,)

    b =
      (,
        [1] = 1
        [2] = 2
      ,)
  }

  octave:4> struct ("a",{1},"b",{1,2},"c",{})
  error: struct: dimensions of parameter 4 do not match those of parameter 6
  octave:4> struct ()

  *** struct:

  struct('field',value,'field',value,...)

  Create a structure and initialize its value.

  struct('field',{values},'field',{values},...)

  Create a structure array and initialize its values.  The dimensions
  of each array of values must match.  Singleton cells and non-cell values
  are repeated so that they fill the entire array.

  struct('field',{},'field',{},...)

Create an empty structure array.
======================================================================

  Once upon a time, struct() returned an empty struct, without need to
do struct([]). No field-value pair -> empty struct, right?

  I don't recall that there's anything wrong with having struct values
being lists, my code has relied on this for some time now.

  Any tips welcome,

  Etienne

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

[OctDev] sparse: size and indexing

[Schloegl A. <alo...@tu...>]

I'm using the latest CVS-versions of Octave and Octave-forge and observed
two problems using SPARSE matrices.

(1) size of a sparse matrix can not be obtained, [-1,-1] is returned.

(2) indexing of a sparse field within a struct stops with an error.

Issue (1) was working well in Octave4Windows 2.1.50. Obviously, it broke at some
time


Repeat-By:
---------

 octave:1> which sparse
sparse is the dynamically-linked function from the file
/usr/local/libexec/octave/2.1.57/site/oct/i686-pc-linux-gnu/octave-forge/sparse.oct
octave:2>  x=sparse(2:31,1:30,1,31,30);
octave:3> size(x)
ans =

  -1  -1

octave:4>  H.x=sparse(2:31,1:30,1,31,30);
octave:5> size(H.x)
ans =

  -1  -1

octave:6> x(2:4,:)    % Here, indexing works well
ans =

Compressed Column Sparse (rows=3, cols=30, nnz=3)
  (1 , 1) -> 1
  (2 , 2) -> 1
  (3 , 3) -> 1

octave:7> H.x(2:4,:)   % Here, indexing does not work
error: can't perform indexing operations for sparse type
octave:7>