Ben Abbott | 1 Sep 03:51 2008
Picon

seg-fault for default branch on Mac OSX?

Does anyone have a thought as to what might cause this?

octave:1> backend('fltk')
octave:2> clear -f
octave:3> panic: Segmentation fault -- stopping myself...
attempting to save variables to `octave-core'...
save to `octave-core' complete
Segmentation fault

Ben

Ben Abbott | 1 Sep 04:51 2008
Picon

Re: seg-fault for default branch on Mac OSX?


On Aug 31, 2008, at 9:51 PM, Ben Abbott wrote:

> Does anyone have a thought as to what might cause this?
>
> octave:1> backend('fltk')
> octave:2> clear -f
> octave:3> panic: Segmentation fault -- stopping myself...
> attempting to save variables to `octave-core'...
> save to `octave-core' complete
> Segmentation fault
>
> Ben

I've done some more isolating

octave:1> __init_fltk__
octave:2> clear -f
octave:3> panic: Segmentation fault -- stopping myself...
attempting to save variables to `octave-core'...
save to `octave-core' complete

Ben

Ben Abbott | 1 Sep 05:15 2008
Picon

Re: seg-fault for default branch on Mac OSX?


On Aug 31, 2008, at 9:51 PM, Ben Abbott wrote:

> Does anyone have a thought as to what might cause this?
>
> octave:1> backend('fltk')
> octave:2> clear -f
> octave:3> panic: Segmentation fault -- stopping myself...
> attempting to save variables to `octave-core'...
> save to `octave-core' complete
> Segmentation fault
>
> Ben

The following produces no ill effect.

octave:1> __init_fltk__
octave:2> __remove_fltk__
octave:3> clear -f

Ben

Michael Goffioul | 1 Sep 09:04 2008
Picon

Re: seg-fault for default branch on Mac OSX?

On Mon, Sep 1, 2008 at 5:15 AM, Ben Abbott <bpabbott <at> mac.com> wrote:
> The following produces no ill effect.
>
> octave:1> __init_fltk__
> octave:2> __remove_fltk__
> octave:3> clear -f

Most probably due to a missing lock in FLTK backend
code. Can you try if adding "mlock()" in __init_fltk__
code in fltk_backend.cc does fix the problem? If yes,
I'll propose a complete patch later today.

Michael.

Michael Goffioul | 1 Sep 10:04 2008
Picon

Question about shipping OpenSSL with octave binaries?

Hi,

I would like to know if anybody has some hints about licensing
problems regarding shipping OpenSSL with the octave binary
package. I've read various things about possible incompatibilities
between OpenSSL license and GPL [1], and I'm a little bit confused.

Michael.

[1] http://www.openssl.org/support/faq.html#LEGAL2

Thomas Weber | 1 Sep 10:24 2008
Picon

Re: Question about shipping OpenSSL with octave binaries?

Am Montag, den 01.09.2008, 10:04 +0200 schrieb Michael Goffioul:
> Hi,
> 
> I would like to know if anybody has some hints about licensing
> problems regarding shipping OpenSSL with the octave binary
> package. I've read various things about possible incompatibilities
> between OpenSSL license and GPL [1], and I'm a little bit confused.

The best description I know of is
http://www.gnome.org/~markmc/openssl-and-the-gpl.html

The usual way around this is to add a "link exception" to the GPL'd
code. Probably won't work for Octave due to the sheer number of
copyright holders.

What do you need OpenSSL for? Maybe GnuTLS is the easiest way out.

	Thomas

Michael Goffioul | 1 Sep 10:34 2008
Picon

Re: Question about shipping OpenSSL with octave binaries?

On Mon, Sep 1, 2008 at 10:24 AM, Thomas Weber
<thomas.weber.mail <at> gmail.com> wrote:
> Am Montag, den 01.09.2008, 10:04 +0200 schrieb Michael Goffioul:
>> Hi,
>>
>> I would like to know if anybody has some hints about licensing
>> problems regarding shipping OpenSSL with the octave binary
>> package. I've read various things about possible incompatibilities
>> between OpenSSL license and GPL [1], and I'm a little bit confused.
>
> The best description I know of is
> http://www.gnome.org/~markmc/openssl-and-the-gpl.html
>
> The usual way around this is to add a "link exception" to the GPL'd
> code. Probably won't work for Octave due to the sheer number of
> copyright holders.
>
> What do you need OpenSSL for? Maybe GnuTLS is the easiest way out.

I don't personally "need" OpenSSL. But to make urlwrite more complete,
providing a cURL compiled against OpenSSL is probably better. I already
had a look at GnuTLS, but OpenSSL compiles under MSVC out of the box,
while GnuTLS does not (and it's not straightforward to make it work).

Michael.

David Bateman | 1 Sep 11:44 2008

Re: return type of surfc for MatlabR2008a or later


> a1 = 
>
>     'surface'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>     'patch'
>
>   
This is the same output as earlier versions of Matlab. Are you sure this 
is Matlab 2008a? If so it means that the surfc plotting command doesn't 
return a group object like all of the other commands and that therefore 
it won't work with the new linkdata command of matlab. I'd consider this 
a bug in matlab if this was 2008a...

Ok, assuming that this is the output of 2008a, I'd implement it how it 
makes sense to me and fix it later when matlab gets their act together.

D.

--

-- 
(Continue reading)

David Bateman | 1 Sep 12:28 2008

Re: [changeset] property value not paired with name in __add_datasource__.m

Ben Abbott wrote:
>
> On Aug 31, 2008, at 12:52 PM, Ben Abbott wrote:
>
>>
>> On Aug 31, 2008, at 11:15 AM, John W. Eaton wrote:
>>
>>> On 31-Aug-2008, Michael Goffioul wrote:
>>>
>>> | The following patch (in graphics mq) solves the problem
>>> | 
>>> http://hg.tw-math.de/octave-graphics-mq/file/9d2c6e36870f/set_empty_args 
>>>
>>>
>>> I applied it.
>>>
>>> Thanks,
>>>
>>> jwe
>>
>> I started fresh, but the default branch still fails to build ... new 
>> error though :-)
>>
>> ../../run-octave -f -q -H -p . --eval "geometryimages ('voronoi', 
>> 'eps');"
>> error: set: invalid number of arguments
>> error: called from:
>> error:   
>> /Users/bpabbott/Development/mercurial/octave-3-0-0/scripts/plot/__plt2vv__.m 
>> at line 83, column 7
(Continue reading)

David Bateman | 1 Sep 12:30 2008

Re: [changeset] property value not paired with name in __add_datasource__.m

Ben Abbott wrote:
>
> On Aug 31, 2008, at 11:15 AM, John W. Eaton wrote:
>
>> On 31-Aug-2008, Michael Goffioul wrote:
>>
>> | The following patch (in graphics mq) solves the problem
>> | 
>> http://hg.tw-math.de/octave-graphics-mq/file/9d2c6e36870f/set_empty_args
>>
>> I applied it.
>>
>> Thanks,
>>
>> jwe
>
> I started fresh, but the default branch still fails to build ... new 
> error though :-)
>
> ../../run-octave -f -q -H -p . --eval "geometryimages ('voronoi', 
> 'eps');"
> error: set: invalid number of arguments
> error: called from:
> error:   
> /Users/bpabbott/Development/mercurial/octave-3-0-0/scripts/plot/__plt2vv__.m 
> at line 83, column 7
> error:   
> /Users/bpabbott/Development/mercurial/octave-3-0-0/scripts/plot/__plt2__.m 
> at line 68, column 14
> error:   
(Continue reading)


Gmane