headers
John W. Eaton | 1 Sep 2004 16:31
Picon

MATLAB load/save changes

On 18-Aug-2004, Quentin Spencer <qspencer <at> ieee.org> wrote:

| I'm not sure whether to report this as a bug or just as an FYI to the 
| maintainers, so I'm posting it to both. Back in March, John sent out a 
| list of new features listed as new in the then-upcoming MATLAB 7. I've 
| just run into a load/save incompatibility that wasn't in that list. 
| Trying to load a file saved in MATLAB 7, I got an error. In the 
| documentation for MATLAB's save function, I found the following:
| 
|     By default, MAT-files created with SAVE are compressed and char 
| arrays are
|     encoded using Unicode. These MAT-files cannot be loaded into versions of
|     MATLAB prior to MATLAB 7.0. The -V6 option disables these features and
|     allows saved MAT-files to load into older versions of MATLAB. To disable
|     these features by default, modify the settings in the General->MAT-Files
|     preferences panel, accessible via the File->Preferences menu item. With
|     compression enabled, saving data that does not compress well takes
|     longer. In this case, the -V6 option may be preferable.
| 
| The file in question didn't have any char arrays, so I suspect it was 
| the compression that tripped it up. I looked briefly at the file format 
| document on their web site:
|     
| http://www.mathworks.com/access/helpdesk/help/pdf_doc/matlab/matfile_format.pdf
| It looks like it's all documented--they are using gzip compression on 
| each variable.

Thanks for the pointer.  I think Octave should handle the new file
format, but it is a low priority project for me right now.  If someone
else can contribute the code, that would be helpful.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Hall, Benjamin | 1 Sep 2004 16:43

RE: MATLAB load/save changes

>> a = int8(5)/int8(3)

a =

    2

>> whos
  Name      Size                    Bytes  Class

  a         1x1                         1  int8 array
  ans       1x1                         1  int8 array

Grand total is 2 elements using 2 bytes

-----Original Message-----
From: John W. Eaton [mailto:jwe <at> bevo.che.wisc.edu]
Sent: Wednesday, September 01, 2004 10:31 AM
To: Quentin Spencer
Cc: maintainers <at> octave.org
Subject: MATLAB load/save changes

On 18-Aug-2004, Quentin Spencer <qspencer <at> ieee.org> wrote:

| I'm not sure whether to report this as a bug or just as an FYI to the 
| maintainers, so I'm posting it to both. Back in March, John sent out a 
| list of new features listed as new in the then-upcoming MATLAB 7. I've 
| just run into a load/save incompatibility that wasn't in that list. 
| Trying to load a file saved in MATLAB 7, I got an error. In the 
| documentation for MATLAB's save function, I found the following:
|
(Continue reading)

Permalink | Reply | 0 comments |
headers
Przemek Klosowski | 1 Sep 2004 19:30

Re: {Spam?} zeros, ones, eye with integer type args...


This is rather off topic, but you may find it amusing. 
As you can see in the subject, David's recent patch was labeled
as spam by our mail server. Specifically, it  the HOT_NASTY
module gave it a perfect 10.0! I am wondering which words caused it:

       - kron ? 
       - matrix ?
       - single ?
       - commune ?
       - classes ?  :)
       - XXX FIXME XXX ?

Between indecency filters and terrorism filters, one has to watch
one's words... :)
Permalink | Reply | 1 comment |
headers
John W. Eaton | 1 Sep 2004 20:32
Picon

integer artithmetic

What should we do for things like

  int16(0) / int16(0)

or

  Inf * int16(0)

?

If both arguments are double, these operations genrate NaNs.  What
should Octave do, given the constraint that the result should be an
int16 object (at least I believe that's the rule that we have to
follow for compatibility)?

Should the result be int16(0)?

Can someone who has Matlab R14 say what it does?

Thanks,

jwe
Permalink | Reply | 0 comments |
headers
Hall, Benjamin | 1 Sep 2004 20:42

RE: integer artithmetic

Here's R14 results:

>> a = int16(0)/int16(0)
Warning: Divide by zero.

a =

      0

>> b = Inf*int16(0)

b =

      0

>> whos
  Name      Size                    Bytes  Class

  a         1x1                         2  int16 array
  ans       1x1                         2  int16 array
  b         1x1                         2  int16 array

Grand total is 3 elements using 6 bytes

-----Original Message-----
From: John W. Eaton [mailto:jwe <at> bevo.che.wisc.edu]
Sent: Wednesday, September 01, 2004 2:32 PM
To: octave maintainers mailing list
Subject: integer artithmetic
(Continue reading)

Permalink | Reply | 501 comments |
headers
John W. Eaton | 1 Sep 2004 21:03
Picon

RE: integer artithmetic

On  1-Sep-2004, Hall, Benjamin <benjamin.hall <at> pw.utc.com> wrote:

| Here's R14 results:
| 
| >> a = int16(0)/int16(0)
| Warning: Divide by zero.
| 
| a =
| 
|       0
| 
| >> b = Inf*int16(0)
| 
| b =
| 
|       0

OK, this is what I've done for Octave as well.

Thanks,

jwe
Permalink | Reply | 501 comments |
headers
David Bateman | 1 Sep 2004 21:54

Re: {Spam?} zeros, ones, eye with integer type args...

Hey its not often I score a perfect 10....

D.

According to Przemek Klosowski <przemek <at> jazz.ncnr.nist.gov> (on 09/01/04):
> 
> This is rather off topic, but you may find it amusing. 
> As you can see in the subject, David's recent patch was labeled
> as spam by our mail server. Specifically, it  the HOT_NASTY
> module gave it a perfect 10.0! I am wondering which words caused it:
> 
>        - kron ? 
>        - matrix ?
>        - single ?
>        - commune ?
>        - classes ?  :)
>        - XXX FIXME XXX ?
>     
> Between indecency filters and terrorism filters, one has to watch
> one's words... :)

--

-- 
David Bateman                                David.Bateman <at> motorola.com
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
(Continue reading)

Permalink | Reply | 0 comments |
headers
John W. Eaton | 2 Sep 2004 07:31
Picon

Octave 2.1.58 available for ftp

Octave 2.1.58 is now available for ftp from ftp.octave.org in the
directory /pub/octave/bleeding-edge:

  -rw-r--r--  1 1005  5446553 Sep  2 04:06 octave-2.1.58.tar.gz
  -rw-r--r--  1 1005  4249914 Sep  2 04:06 octave-2.1.58.tar.bz2
  -rw-r--r--  1 1005   285012 Sep  2 04:08 octave-2.1.57-2.1.58.patch.gz
  -rw-r--r--  1 1005   225593 Sep  2 04:08 octave-2.1.57-2.1.58.patch.bz2

  5952e00626894997443dfb463ed4bd46  octave-2.1.58.tar.gz
  769a1ff3a27e411ffe3cf8930ea126c2  octave-2.1.58.tar.bz2
  c15c76fd944a99c2354cacd7fbf349db  octave-2.1.57-2.1.58.patch.gz
  dd12e5d39e8d61264661776552592015  octave-2.1.57-2.1.58.patch.bz2

Thanks to David Bateman for all his hard work to get this snapshot
ready.

This version includes many new features, including integer data types,
inline functions, function handles, concatenation of structs, cell
arrays and user-defined types, and better compatibility with the
leading brand.

Although I believe that 2.1.58 will be quite useable, there have been
many changes and experience says that any number of unexpected problems
could show up just after the tar file hits the ftp site, so 2.1.57
remains the recommended version for now.

As always, if your favorite bug is still not fixed, please report it.

jwe
(Continue reading)

Permalink | Reply | 6 comments |
headers
David Bateman | 2 Sep 2004 09:26

Re: Octave 2.1.58 available for ftp


For those who want to build octave-forge with this, you'll need the
octave-forge CVS, particularly for my stuff :-)

D.

According to John W. Eaton <jwe <at> bevo.che.wisc.edu> (on 09/02/04):
> Octave 2.1.58 is now available for ftp from ftp.octave.org in the
> directory /pub/octave/bleeding-edge:
> 
>   -rw-r--r--  1 1005  5446553 Sep  2 04:06 octave-2.1.58.tar.gz
>   -rw-r--r--  1 1005  4249914 Sep  2 04:06 octave-2.1.58.tar.bz2
>   -rw-r--r--  1 1005   285012 Sep  2 04:08 octave-2.1.57-2.1.58.patch.gz
>   -rw-r--r--  1 1005   225593 Sep  2 04:08 octave-2.1.57-2.1.58.patch.bz2
> 
>   5952e00626894997443dfb463ed4bd46  octave-2.1.58.tar.gz
>   769a1ff3a27e411ffe3cf8930ea126c2  octave-2.1.58.tar.bz2
>   c15c76fd944a99c2354cacd7fbf349db  octave-2.1.57-2.1.58.patch.gz
>   dd12e5d39e8d61264661776552592015  octave-2.1.57-2.1.58.patch.bz2
> 
> Thanks to David Bateman for all his hard work to get this snapshot
> ready.
> 
> This version includes many new features, including integer data types,
> inline functions, function handles, concatenation of structs, cell
> arrays and user-defined types, and better compatibility with the
> leading brand.
> 
> Although I believe that 2.1.58 will be quite useable, there have been
> many changes and experience says that any number of unexpected problems
(Continue reading)

Permalink | Reply | 5 comments |
headers
Przemek Klosowski | 2 Sep 2004 21:40

Re: Octave 2.1.58 available for ftp


With the latest octave (2.1.58), 'a=1; a(2)=2' returns 'a=1 2', as
always; but 'b=int8(1);b(2)=2' results in an error:

  operator =: no conversion for assignment of 'scalar' to indexed 'int8 scalar'

The similar error happens for int16 and int32.
Permalink | Reply | 2 comments |

Gmane