headers
Lippert, Ross A. | 17 Jun 2003 16:46

dld files in 2.1.48


Has anyone run into this?

octave:1> oregonator
error: /home/ripper/m/oregonator.oct is not a valid shared library

Is it possible that I could get this far and not have shared library
support
compiled in?  Could that be it?

I have re-done 'mkoctfile oregonator.cc' and it is still there.

-r

-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------
Permalink | Reply | 0 comments |
headers
Pascal A. Dupuis | 17 Jun 2003 17:33
Picon
Picon

Change if fscanf ?

To: bug-octave <at> bevo.che.wisc.edu
Cc: 
Subject: Change if fscanf ?

Bug report for Octave 2.1.49 configured for i386-pc-linux-gnu

Description:
-----------

  * Was there recently a change in fscanf? I try to read a text file, where
    some values are 'Inf' or '-Inf'. With 2.1.36 I got numerical plus or
    minus infinite, and now it seems to stop the fscanf process ? The 
    workaround is to read a data file line by line, and process each line
    in turn, but this is much slower.

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

  * Given the following file: ('data.txt')
	0	1
	1	Inf
	2	-Inf
	3	+Inf
	4       2
  * f = fopen('data.txt', 'r')
    [data, count] = fscanf(f, '%f ')
    data should be [ 0 1 1 Inf 2 -Inf 3 Inf 4 2].', and count 10. Here 
    data is truncated as [ 0 1 1].', and count is 3   
Fix:
---
(Continue reading)

Permalink | Reply | 1 comment |
headers
John W. Eaton | 17 Jun 2003 17:45
Picon

Change if fscanf ?

On 17-Jun-2003, Pascal A. Dupuis <Pascal.Dupuis <at> esat.kuleuven.ac.be> wrote:

| To: bug-octave <at> bevo.che.wisc.edu
| Cc: 
| Subject: Change if fscanf ?
| 
| Bug report for Octave 2.1.49 configured for i386-pc-linux-gnu
| 
| Description:
| -----------
| 
|   * Was there recently a change in fscanf? I try to read a text file, where
|     some values are 'Inf' or '-Inf'. With 2.1.36 I got numerical plus or
|     minus infinite, and now it seems to stop the fscanf process ? The 
|     workaround is to read a data file line by line, and process each line
|     in turn, but this is much slower.
| 
| Repeat-By:
| ---------
| 
|   * Given the following file: ('data.txt')
| 	0	1
| 	1	Inf
| 	2	-Inf
| 	3	+Inf
| 	4       2
|   * f = fopen('data.txt', 'r')
|     [data, count] = fscanf(f, '%f ')
|     data should be [ 0 1 1 Inf 2 -Inf 3 Inf 4 2].', and count 10. Here 
|     data is truncated as [ 0 1 1].', and count is 3
(Continue reading)

Permalink | Reply | 0 comments |
headers
John W. Eaton | 17 Jun 2003 17:57
Picon

Panic on cygwin w/ ctrl-up arrow

On 16-Jun-2003, Sharp, Gregory C. <GCSHARP <at> PARTNERS.ORG> wrote:

| I found that cygwin octave crashes when I press CTRL+UP ARROW.
| It's hard to see the message, but seems to be a panic with signal 11.
| Crash is ONLY for the rxvt version.  The cmd version works fine.
| 
| I am using the version 2.1.14 binary distribution, no-atlas,
| without a separate cygwin installation, on windows 2000.

I don't see this problem with the latest version or rxvt (I installed
it a few minutes ago) and the latest Octave 2.1.49 on a Windows 2000
system.

There is a binary distribution of 2.1.49 for Cygwin available from
ftp.octave.org.  Directions for installing it can be found here:

  http://www.octave.org/mailing-lists/help-octave/2003/990

If you still have this problem with the latest Octave and rxvt, can
you please post more details?  Do you start Octave or rxvt with any
options?

Thanks,

jwe

-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
(Continue reading)

Permalink | Reply | 0 comments |
headers
Aaron A. King | 17 Jun 2003 18:33
Picon

Unidentified subject!

To: bug-octave <at> bevo.che.wisc.edu
Cc: king <at> tiem.utk.edu
Subject: bug in hist.m

Bug report for Octave 2.1.49 configured for i686-pc-linux-gnu

Description:
-----------

Bug caused by small oversight in function hist (plot/hist.m).  The bug
only occurs when the 2nd argument is a vector.  Easily fixed (patch
included).

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

data = 10+5*randn(10000,1);
bins = -10:30;
[n,y] = hist(data, bins);

Fix:
---

The local variable "n" is not defined in this case.  The correct
definition is obvious (see patch).

---BEGIN PATCH--------------------------
*** /usr/local/share/octave/2.1.49/m/plot/hist.m	2003-06-05 15:45:46.000000000 -0400
--- hist.m	2003-06-17 11:18:52.000000000 -0400
*************** function [nn, xx] = hist (y, x, norm)
(Continue reading)

Permalink | Reply | 1 comment |
headers
John W. Eaton | 17 Jun 2003 18:50
Picon

Unidentified subject!

On 17-Jun-2003, Aaron A. King <king <at> quercus.tiem.utk.edu> wrote:

| To: bug-octave <at> bevo.che.wisc.edu
| Cc: king <at> tiem.utk.edu
| Subject: bug in hist.m
| 
| Bug report for Octave 2.1.49 configured for i686-pc-linux-gnu
| 
| Description:
| -----------
| 
| Bug caused by small oversight in function hist (plot/hist.m).  The bug
| only occurs when the 2nd argument is a vector.  Easily fixed (patch
| included).
| 
| Repeat-By:
| ---------
| 
| data = 10+5*randn(10000,1);
| bins = -10:30;
| [n,y] = hist(data, bins);
| 
| Fix:
| ---
| 
| The local variable "n" is not defined in this case.  The correct
| definition is obvious (see patch).

I applied your patch.
(Continue reading)

Permalink | Reply | 0 comments |
headers
David Witherspoon | 17 Jun 2003 19:17
Picon

Unidentified subject!

To: bug-octave <at> bevo.che.wisc.edu
Cc: djw
Subject: bartlett_test uses wrong d.f.

Bug report for Octave 2.1.36 configured for i686-pc-linux-gnu

Description:
-----------

It appears that the octave function bartlett_test uses k degrees of freedom (k = number of sets of
measurements) instead of k-1 to calculate the returned p-value from a chi-squared distribution.  I've
checked several references, they all use k-1 df.

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

Evident in the source code;  also, compute Bartlett's test by hand, compare to results from octave.

Fix:
---

replace 
df = k;
with 
df = k - 1;

Configuration (please do not edit this section):
-----------------------------------------------

uname output:     Linux shiva 2.4.20-18.8smp #1 SMP Thu May 29 07:20:32 EDT 2003 i686 i686 i386 GNU/Linux
(Continue reading)

Permalink | Reply | 0 comments |
headers
Gregory Sharp | 18 Jun 2003 04:01
Picon
Favicon

Re: FW: Panic on cygwin w/ ctrl-up arrow

> | I found that cygwin octave crashes when I press CTRL+UP
> ARROW.
> | It's hard to see the message, but seems to be a panic with
> signal 11.
> | Crash is ONLY for the rxvt version.  The cmd version works
> fine.
> | 
> | I am using the version 2.1.14 binary distribution, no-atlas,
> | without a separate cygwin installation, on windows 2000.
> 
> I don't see this problem with the latest version or rxvt (I
> installed
> it a few minutes ago) and the latest Octave 2.1.49 on a
> Windows 2000
> system.

Hi,
I'm sorry, I didn't know my version was old.  I'll repost 
if I find the problem in the newer version.

By the way, the windows FAQ suggests using 2.1.42 binary 
from the sourceforge site.

Thanks,
Greg Sharp
gregsharp <at> geocities.com

__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
(Continue reading)

Permalink | Reply | 0 comments |
headers
Olli Saarela | 19 Jun 2003 16:48
Picon

Re: Off-by-one problem with date numbers

>      >> datestr(datenum(2003,6,5,13,0,0), 0)
>      ans = 04-Jun-2003 12:59:59
>
> Problem 1: date is off by one
>
> Problem 2: a small round-off error in time
>

The following is my attempt at handling the round-off error in Problem
2. Any comments or suggestions?

 Olli

--- datevec.m  Thu Jun 19 17:22:33 2003
+++ datevec.m   Thu Jun 19 17:22:19 2003
 <at>  <at>  -74,6 +74,14  <at>  <at> 
        s(i) = t;
       else
        tm = gmtime((date(i) - 719529)*86400);
+       % Prettier rounding for integer number of seconds.
+       date2 = date(i) * (1 + 3*eps);
+       tm2 = gmtime((date2 - 719529)*86400);
+       if (tm.sec ~= tm2.sec)
+         tm = tm2;
+         tm.usec = 0;
+       end
+       %
        Y(i) = tm.year+1900;
        M(i) = tm.mon+1;
        D(i) = tm.mday;
(Continue reading)

Permalink | Reply | 0 comments |
headers
John W. Eaton | 19 Jun 2003 18:40
Picon

'end' as last index for structs

On  5-Mar-2003, I wrote:

| On  5-Mar-2003, Mats Jansson <mats.e.jansson <at> home.se> wrote:
| 
| | Bug report for Octave 2.1.46 configured for i686-pc-linux-gnu
| 
| | octave:1> a.a = 1:5
| | a =
| | {
| |   a =
| | 
| |     1  2  3  4  5
| | 
| | }
| | 
| | octave:2> a.a(3:end)
| | ans = [](0x0)
| | octave:3> (a.a)(3:end)
| | ans = [](0x0)
| | octave:4> b = a.a; b(3:end)
| | ans =
| | 
| |   3  4  5
| 
| I see the problem, but no easy fix.  I think it may require major
| surgery.

It turns out that the surgery was not that major, but it was somewhat
tricky to figure out how do it without a making a big mess...
(Continue reading)

Permalink | Reply | 0 comments |

Gmane