hl2modrift | 20 Dec 03:18 2014
Picon

Compiled Octave 3.8.2 for Android. Plotting with gnuplot?

I've compiled Octave 3.8.2 and all of its' dependencies, as well as most of
its' optional dependencies, minus hdf5 and Graphics Magick, for Android.
Specifically, to run on an HTC One m8.

Here's what I used for configure:

--host=arm-cortexa9_neon-linux-gnueabihf
--prefix=/home/shadowed/octave/native --disable-java --disable-gui
--without-hdf5 --with-x=no --without-framework-carbon --without-opengl
--without-framework-opengl --with-sysroot=/home/shadowed/octave/native
--with-blas=/home/shadowed/octave/native/lib/libblas.a
--with-lapack=/home/shadowed/octave/native/lib/liblapack.a
host_alias=arm-cortexa9_neon-linux-gnueabihf
LDFLAGS=-L/home/shadowed/octave/native/lib
CPPFLAGS=-I/home/shadowed/octave/native/include

It is nearly flawless. Nearly. I can't plot, or, more accurately, I can't
set the default plotter.

As far as I can tell, none of the options I've selected keep the finished
binary from plotting with gnuplot, which I have set up to create a .png
image (/sdcard2/plot.png). However, when I run:

        octave: 1> graphics_toolkit("gnuplot")

It returns:

        error: 'graphics_toolkit' undefined near line 1 column 1

I can register gnuplot just fine, and it will show in the list of available
(Continue reading)

Forrest Carroll | 20 Dec 05:30 2014
Picon

Compiled Octave 3.8.2 for Android. Plotting with gnuplot?

I've compiled Octave 3.8.2 and all of its' dependencies, as well as most of its' optional dependencies, minus hdf5 and Graphics Magick, for Android. Specifically, to run on an HTC One m8.

Here's what I used for configure:

--host=arm-cortexa9_neon-linux-gnueabihf --prefix=/home/shadowed/octave/native --disable-java --disable-gui --without-hdf5 --with-x=no --without-framework-carbon --without-opengl --without-framework-opengl --with-sysroot=/home/shadowed/octave/native --with-blas=/home/shadowed/octave/native/lib/libblas.a --with-lapack=/home/shadowed/octave/native/lib/liblapack.a host_alias=arm-cortexa9_neon-linux-gnueabihf LDFLAGS=-L/home/shadowed/octave/native/lib CPPFLAGS=-I/home/shadowed/octave/native/include

It is nearly flawless. Nearly. I can't plot, or, more accurately, I can't set the default plotter.

As far as I can tell, none of the options I've selected keep the finished binary from plotting with gnuplot, which I have set up to create a .png image (/sdcard2/plot.png). However, when I run:

        octave: 1> graphics_toolkit("gnuplot")

It returns:

        error: 'graphics_toolkit' undefined near line 1 column 1

I can register gnuplot just fine, and it will show in the list of available graphics toolkits, but Octave refuses to load it.

Similar errors I've seen around are related to old binaries, but this is the last release. Does anyone have an idea of how to fix this? Recompiling is not an issue, if need be, but I'd like to know what might fix it beforehand.

Thanks.

-FC
Rik | 19 Dec 18:31 2014

Re: WTF bug ID

On 12/19/2014 09:00 AM, octave-maintainers-request <at> gnu.org wrote:
Subject:
New Item Group value for bug tracker
From:
"John W. Eaton" <jwe <at> octave.org>
Date:
12/19/2014 07:27 AM
To:
Octave-maintainers <at> octave.org
List-Post:
<mailto:octave-maintainers <at> gnu.org>
Content-Transfer-Encoding:
7bit
Precedence:
list
MIME-Version:
1.0
Message-ID:
<549443FE.60808 <at> octave.org>
Content-Type:
text/plain; charset=utf-8; format=flowed
Message:
6

While thinking about warning and error IDs, I came across the following Matlab behavior:

Escape sequences like \n and \t are not converted in error messages if a single argument is passed to error, but they are if more than one argument is supplied AND the message ID counts as the second argument, not just the ones that follow as possible values for printf-style % conversions.  Weird.  OK, so I was going to note this in the bug tracker, and I started to think about how crazy this behavior seems to me, so I decided to add a new Item Group value to the tracker:

  WTF, MATLAB?!?  -- Insane Matlab bugs^H^H^H^Hfeatures we'd really prefer not to copy

Then, almost finished with my report, I decided to try this in Octave. Much to my surprise, it appears that we've already copied this behavior.  WTF, Octave?!?

So, anyway, I left the value in the tracker.  Feel free to enjoy using it anytime someone reports some crazy Matlab compatibility issue.

jwe

I remember being unhappy about having that implemented at the time, but whatever.  I had added a note at the bottom of the docstring for error to try and warn people about this ("two or more input arguments")

--Rik

Implementation Note: For compatibility with MATLAB, escape
     sequences (e.g., "\n" => newline) are processed in TEMPLATE
     regardless of whether TEMPLATE has been defined within single
     quotes as long as there are two or more input arguments.  Use a
     second backslash to stop interpolation of the escape sequence
     (e.g., "\\n") or use the `regexptranslate' function.


John W. Eaton | 19 Dec 16:27 2014

New Item Group value for bug tracker

While thinking about warning and error IDs, I came across the following 
Matlab behavior:

Escape sequences like \n and \t are not converted in error messages if a 
single argument is passed to error, but they are if more than one 
argument is supplied AND the message ID counts as the second argument, 
not just the ones that follow as possible values for printf-style % 
conversions.  Weird.  OK, so I was going to note this in the bug 
tracker, and I started to think about how crazy this behavior seems to 
me, so I decided to add a new Item Group value to the tracker:

   WTF, MATLAB?!?  -- Insane Matlab bugs^H^H^H^Hfeatures we'd really 
prefer not to copy

Then, almost finished with my report, I decided to try this in Octave. 
Much to my surprise, it appears that we've already copied this behavior. 
  WTF, Octave?!?

So, anyway, I left the value in the tracker.  Feel free to enjoy using 
it anytime someone reports some crazy Matlab compatibility issue.

jwe

Doug Stewart | 18 Dec 21:21 2014
Picon

GUI editor

I have been using octave 4.1.0+ . I find that when you shut down octave, but have an unsaved file in the editor, then an option box pops up and you are given two options.

1) Close without Saving
2) Save

While this Box is showing, you cannot do anything else in Octave.

I would like the option of seeing the unsaved file --- to decide if it really needs saving, or not.

One way to do this would be to add a 3rd option.
3) Cancel

This would put you back in a running octave environment and allow you to keep working, or look at the files, etc.


Doug Stewart

PS Thanks to all who have worked on the GUI.
--
DAS

Rik | 18 Dec 19:39 2014

Re: style for error and warning messages

On 12/18/2014 09:00 AM, octave-maintainers-request <at> gnu.org wrote:
Subject:
style for error and warning messages
From:
"John W. Eaton" <jwe <at> octave.org>
Date:
12/18/2014 08:45 AM
To:
Octave-maintainers <at> octave.org
List-Post:
<mailto:octave-maintainers <at> gnu.org>
Content-Transfer-Encoding:
7bit
Precedence:
list
MIME-Version:
1.0
Message-ID:
<549304C7.70208 <at> octave.org>
Content-Type:
text/plain; charset=utf-8; format=flowed
Message:
3

Currently, most of Octave uses a style for error and warning messages that is something like this:

  warning ("nchoosek: possible loss of precision");

but now that error and warning messages display stack trace information by default, having the name of the function as a prefix for the message is redundant.  Does anyone object to removing the function name as a prefix to error and warning messages?


Seems reasonable.

--Rik
John W. Eaton | 18 Dec 17:45 2014

style for error and warning messages

Currently, most of Octave uses a style for error and warning messages 
that is something like this:

   warning ("nchoosek: possible loss of precision");

but now that error and warning messages display stack trace information 
by default, having the name of the function as a prefix for the message 
is redundant.  Does anyone object to removing the function name as a 
prefix to error and warning messages?

jwe

Carlo De Falco | 18 Dec 12:32 2014
Picon

modifications to OF web pages

Hi Carnë,

I see you are making changes to the OF web pages,
which is a good idea as they can use some renovation.

Here are some comments you may or may not find useful:

1) http://octave.sourceforge.net/developers.html

* The first sentence does not sound very clear in describing 
  the different purposes of Octave core and Octave-Forge.
  This entry in the FAQ probably serves the same purpose better:
  http://wiki.octave.org/FAQ#What_is_Octave-Forge.3F

* "see it's page." should probably read "see its page."

* This page used to contain a description of the package
directory structure and of the syntax of the DESCRIPTION
and INDEX files, which you seem to have removed, it would
be nicer to at least leave in place a link to places where
the same information can now be found in the manual.

* "you should mention it on the maintainers mailing list."
  would be nice if this test actually linked to the mailing 
  list subscription page.

* "only older packages, usually unmaintained exist there now." 
   this is simply not true.

HTH,
c.

John W. Eaton | 18 Dec 04:06 2014

Singular matrix warning IDs

I checked in the following change so that warning IDs are used for all 
cases where Octave warns about singular matrices when inverting matrices 
or solving systems of equations.

http://hg.savannah.gnu.org/hgweb/octave/rev/a0c7001cf1a8

Note that in some cases the singular matrix warning is followed by 
another warning message about attempting to find a minimum norm 
solution.  I'm not sure whether that warning should be controlled by a 
separate ID, the same ID as for the singular matrix warning, or simply 
removed.

I'm leaning toward just removing it.  Any objections?

jwe

CdeMills | 17 Dec 18:17 2014
Picon

[ov-java.cc] Introduce a fallback mechanism in case the JVM lib changed

Hello,

There is an long ongoing annoyance in ov-java: the path of libjvm is
statically determined at build time. This means that every version upgrade
requires to rebuild Octave. I experienced it a few days ago on Fedora 21. I
produced a patch introducing a fallback mechanism: if the original file
can't be found, it is searched from the environment variable "JAVA_JVM_LIB".

The patch was introduced against bug 40111 which is a bit old. Could the
maintainers of the java stack have a look at it ?

Regards

Pascal Dupuis

--
View this message in context: http://octave.1599824.n4.nabble.com/ov-java-cc-Introduce-a-fallback-mechanism-in-case-the-JVM-lib-changed-tp4667861.html
Sent from the Octave - Maintainers mailing list archive at Nabble.com.

John W. Eaton | 17 Dec 17:06 2014

Re: Subplot problems

On 12/17/2014 10:50 AM, Pantxo Diribarne wrote:
> Konstantinos,
>
> Thanks for the pointer, reading and understanding the patch will
> undoubtedly help much. BTW I don't think your patch has been reverted
> (or not completely) as "autopos_tag" property is still present in octave
> 3.8.2. Do you have a reference for this revert or is this a guess from
> the current buggy behavior?

The substantial changes to subplot since the change Konstantions posted 
were these:

http://hg.savannah.gnu.org/hgweb/octave/rev/623cce4577b3
http://hg.savannah.gnu.org/hgweb/octave/rev/93edd282a2bc
http://hg.savannah.gnu.org/hgweb/octave/rev/9304514b566c

Is one of these obviously the one that caused the trouble?

Since these were intentional changes, I assume that they fixed some 
other problem, so simply undoing one or more of then is not likely the 
right thing to do, but maybe they will also give some clues.

jwe


Gmane