[Scilab-users] atomsLoad

classic Classic list List threaded Threaded
23 messages Options
12
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

[Scilab-users] atomsLoad

Hi,
  I face problem with atomsLoad. (It's the first time I use atoms!)

1) I installed scilab-6.1.0.bin.linux-x86_64.tar.gz on my $HOME.
2) I added the atoms repository http://atoms.scilab/org because the default one is http://atoms.scilab.org/6.1 and it gives nothing!
3) I installed emd_toolbox and the installed files are on $HOME/.Scilab/scilab-6.1.0/atoms/ ($HOME/.Scilab/scilab-6.1.0 is the SCIHOME variable in Scilab)

So far, all is good. But the next step, atomsLoad("emd_toolbox") does not work. The message is

atomsLoad: An error occurred while loading 'stftb-1.2-1':
        lib: /home/jbaudais/.Scilab/scilab-6.1.0/atoms/stftb/1.2-1/macros/ is not a valid lib path.
at line   346 of function atomsLoad ( /home/jbaudais/scilab-6.1.0/share/scilab/modules/atoms/macros/atomsLoad.sci line 362 )

Note: emd_toolbox needs stftb but if I try atomsLoad("stftb"), the result is the same.

I don't understand because the directory /home/jbaudais/.Scilab/scilab-6.1.0/atoms/stftb/1.2-1/macros/ exists and it contains the "names" file and all the .bin, .sci files.

Any ideas?

--
Jean-Yves
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
Samuel GOUGEON Samuel GOUGEON
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hello Jean-Yves,

Le 18/12/2020 à 18:38, Jean-Yves Baudais a écrit :
Hi,
  I face problem with atomsLoad. (It's the first time I use atoms!)

1) I installed scilab-6.1.0.bin.linux-x86_64.tar.gz on my $HOME.
2) I added the atoms repository http://atoms.scilab/org because the default one is http://atoms.scilab.org/6.1 and it gives nothing!
3) I installed emd_toolbox and the installed files are on $HOME/.Scilab/scilab-6.1.0/atoms/ ($HOME/.Scilab/scilab-6.1.0 is the SCIHOME variable in Scilab)

So far, all is good. But the next step, atomsLoad("emd_toolbox") does not work. The message is

atomsLoad: An error occurred while loading 'stftb-1.2-1':

Hard to help without knowing what you exactly did to install the toolbox, but clearly, this version of the Time Frequency Toolbox has been shipped for Scilab 5.3, and can in no way work with Scilab 6.1!

Its most recent 2.0.2 version has been compiled for Scilab 6.0. So unless you recompile it by your own for Scilab 6.1, you could not use it on Scilab 6.1 (because it includes some hard-coded functions).

Why have you chosen a so old version?

Regards
Samuel


_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hi,

> Hard to help without knowing what you exactly did to install the toolbox,

As I wrote I did
--> atomsRepositoryAdd("http://atoms.scilab.org")
--> atomsInstall("emd_toolbox")
--> atomsLoad("emd_toolbox")

Nothing else.

> but
> clearly, [ https://atoms.scilab.org/toolboxes/stftb/1.2.1 | this
>        version ] of the Time Frequency Toolbox has been shipped for Scilab 5.3, and can
>        in no way work with Scilab 6.1!
>
> Its most recent 2.0.2 version has been compiled for Scilab 6.0. So unless you
> recompile it by your own for Scilab 6.1, you could not use it on Scilab 6.1
> (because it includes some hard-coded functions).

Ok. So the error message "lib: [...] is not a valid lib path" does not mean the path does not exist but the version 1.2-1 cannot be used. Right?

> Why have you chosen a so old version? Regards
> Samuel

Hum... I chosen nothing, I let Scilab functions atomsInstall() and atomsLoad() do the job...

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
Samuel GOUGEON Samuel GOUGEON
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Le 21/12/2020 à 10:16, Jean-Yves Baudais a écrit :
> Hi,
>
>> Hard to help without knowing what you exactly did to install the toolbox,
> As I wrote I did
> --> atomsRepositoryAdd("http://atoms.scilab.org")
> --> atomsInstall("emd_toolbox")
> --> atomsLoad("emd_toolbox")
>
> Nothing else.

Hello Jean-Yves,

I was not aware that the whole portal "http://atoms.scilab.org" can be
successfully targeted without specifying any Scilab X.y version.
So i was assuming that you dowloaded the binary file by hand,  and then
you installed it offline.

> 2) I added the atoms repository http://atoms.scilab/org because the default one is http://atoms.scilab.org/6.1 and it gives nothing!
I guess that by "it gives nothing" you mean that you did not get any
emd_toolbox version installed.
That's expected, since the last published version with binaries for
Linux was the 1.2 version for Scilab 5.5.
So it is filtered out if the target Scilab version is 6.1.

As far as i know, you won't be actually able to install and use the
emd_toolbox with Scilab 6.1.
You have either to use Scilab 5.5 for this, or to port emb_toolbox to
Scilab 6.1 yourself, by updating and recompiling it.

By the way, the behavior of the ATOMS server when no Scilab version is
specified in the repository URL would deserve either being documented,
or fixed. As far as i understand, then the user is completely free and
responsible of what it does. In your case, by default the very first
emd_toolbox version 1.0 has been installed (why not. It could be the
last available one, instead...). Then the automatical choice of the
default version of the dependency (stftb) is more obscure...

Samuel

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hi,

  Thank you for the explanations. They make me the atom philosophy clearer :-)

>> 2) I added the atoms repository http://atoms.scilab/org because the default one
>> is http://atoms.scilab.org/6.1 and it gives nothing!
> I guess that by "it gives nothing" you mean that you did not get any
> emd_toolbox version installed.

Yes, and I understand now it's rather good nothing was installed because the atoms "emd_toolbox" is not compatible with Scilab-6.1.

> By the way, the behavior of the ATOMS server when no Scilab version is
> specified in the repository URL would deserve either being documented,
> or fixed. As far as i understand, then the user is completely free and
> responsible of what it does.

Yes, but I have to learn and I'm more irresponsible to do what I did :-)

> In your case, by default the very first
> emd_toolbox version 1.0 has been installed (why not. It could be the
> last available one, instead...). Then the automatical choice of the
> default version of the dependency (stftb) is more obscure...

So, I think the best way it to use older Scilab version because
- only too old emd_toolbox is "atomsInstalled" with Scilab-6.1 and not "atomsLoadable"
- with http://atoms.scilab.org/5.3 repository, the atomsLoad() function alway gives "bad library path"
- I'm not able to start with library source code

Two questions:
- Which Scilab version should I used? (only 5.5.2 or 5.1.1 are available on http://www.scilab.org)
- Is the message "bad libray path" given by atomsLoad() mean there is a compatiliby issues? If yes, why does atomsInstall() intall without any error while atomsLoad() give errors?

Thank for all,

--
Jean-Yves
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
Samuel GOUGEON Samuel GOUGEON
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Le 21/12/2020 à 15:53, Jean-Yves Baudais a écrit :
> So, I think the best way it to use older Scilab version because
> - only too old emd_toolbox is "atomsInstalled" with Scilab-6.1 and not "atomsLoadable"
> - with http://atoms.scilab.org/5.3 repository, the atomsLoad() function alway gives "bad library path"
> - I'm not able to start with library source code
>
> Two questions:
> - Which Scilab version should I used? (only 5.5.2 or 5.1.1 are available on http://www.scilab.org)

You must use Scilab 5.5.2 and install the 1.2 version of the toolbox. It
should run out of the box.


_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hi,

> You must use Scilab 5.5.2 and install the 1.2 version of the toolbox. It
> should run out of the box.

Ok. I installed it. "scilab" command and "scilab -nw" or "-nb" or "-ns" or "-noatomsautoload" give the so funny "Segmentation fault" :-( Hopefully, "scilab -nwni" works, but I only have the head_comments help and no plot. So, not so funny, but I can juggle between scilab-5.5.2 and scilab-6.1.0

"scilab -debug" does not work anymore (but I not able to use gdb!)

gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by gdb)
gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by /lib/x86_64-linux-gnu/libreadline.so.7)
[...]
gdb: relocation error: /lib/x86_64-linux-gnu/libncursesw.so.5: symbol _nc_putchar version NCURSES_TINFO_5.9.20150530 not defined in file libtinfo.so.5 with link time reference

and "scilab -nogui" gives (maybe more informative)

Scilab startup function detected that the function proposed to the engine is the wrong one. Usually, it comes from a linker problem in your distribution/OS.
Here, Scilab should have 'libscijvm-disable' defined but gets 'libscijvm' instead.
If you do not know what it means, please report a bug on http://bugzilla.scilab.org/. If you do, you probably know that you should change the link order in SCI/modules/Makefile.am

Is there a simple way to solve this JVM problem? (Note that the default scilab-6.0 installed in my Unbuntu has a JVM problem also and does not work. Maybe I will try to solve it another time.)

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
Clément David-3 Clément David-3
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hello,

It looks like the scilab 5.5.2 binary is not compatible with your system (nc_putchar version error). IMHO it should be better to use the latest Scilab version and try to build dependencies by hand ; regarding the toolboxes:
 * http://forge.scilab.org/index.php/p/stftb/ is not published for 6.1 but might compile on it ; get the sources and try to build on your own machine.
 * https://atoms.scilab.org/toolboxes/emd_toolbox/ already uses the pvApiCtx symbol ; it might be possible to port it to scilab 6.1 without a big refactoring.

Thanks,

Clément

De : users <[hidden email]> de la part de Jean-Yves Baudais <[hidden email]>
Envoyé : mardi 22 décembre 2020 12:36
À : Users mailing list for Scilab <[hidden email]>
Objet : Re: [Scilab-users] atomsLoad
 
Hi,

> You must use Scilab 5.5.2 and install the 1.2 version of the toolbox. It
> should run out of the box.

Ok. I installed it. "scilab" command and "scilab -nw" or "-nb" or "-ns" or "-noatomsautoload" give the so funny "Segmentation fault" :-( Hopefully, "scilab -nwni" works, but I only have the head_comments help and no plot. So, not so funny, but I can juggle between scilab-5.5.2 and scilab-6.1.0

"scilab -debug" does not work anymore (but I not able to use gdb!)

gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by gdb)
gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by /lib/x86_64-linux-gnu/libreadline.so.7)
[...]
gdb: relocation error: /lib/x86_64-linux-gnu/libncursesw.so.5: symbol _nc_putchar version NCURSES_TINFO_5.9.20150530 not defined in file libtinfo.so.5 with link time reference

and "scilab -nogui" gives (maybe more informative)

Scilab startup function detected that the function proposed to the engine is the wrong one. Usually, it comes from a linker problem in your distribution/OS.
Here, Scilab should have 'libscijvm-disable' defined but gets 'libscijvm' instead.
If you do not know what it means, please report a bug on http://bugzilla.scilab.org/. If you do, you probably know that you should change the link order in SCI/modules/Makefile.am

Is there a simple way to solve this JVM problem? (Note that the default scilab-6.0 installed in my Unbuntu has a JVM problem also and does not work. Maybe I will try to solve it another time.)

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
mottelet mottelet
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hi,


Le 22/12/2020 à 14:47, Clément David a écrit :
Hello,

It looks like the scilab 5.5.2 binary is not compatible with your system (nc_putchar version error). IMHO it should be better to use the latest Scilab version and try to build dependencies by hand ; regarding the toolboxes:
 * http://forge.scilab.org/index.php/p/stftb/ is not published for 6.1 but might compile on it ; get the sources and try to build on your own machine.
 * https://atoms.scilab.org/toolboxes/emd_toolbox/ already uses the pvApiCtx symbol ; it might be possible to port it to scilab 6.1 without a big refactoring.

It is quite easy (a 10 min. job). I made the needed modifications and it compiles and loads under OSX. If you give me the rights on Gitlab I can commit the modifications this afternoon.


S.


Thanks,

Clément

De : users [hidden email] de la part de Jean-Yves Baudais [hidden email]
Envoyé : mardi 22 décembre 2020 12:36
À : Users mailing list for Scilab [hidden email]
Objet : Re: [Scilab-users] atomsLoad
 
Hi,

> You must use Scilab 5.5.2 and install the 1.2 version of the toolbox. It
> should run out of the box.

Ok. I installed it. "scilab" command and "scilab -nw" or "-nb" or "-ns" or "-noatomsautoload" give the so funny "Segmentation fault" :-( Hopefully, "scilab -nwni" works, but I only have the head_comments help and no plot. So, not so funny, but I can juggle between scilab-5.5.2 and scilab-6.1.0

"scilab -debug" does not work anymore (but I not able to use gdb!)

gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by gdb)
gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by /lib/x86_64-linux-gnu/libreadline.so.7)
[...]
gdb: relocation error: /lib/x86_64-linux-gnu/libncursesw.so.5: symbol _nc_putchar version NCURSES_TINFO_5.9.20150530 not defined in file libtinfo.so.5 with link time reference

and "scilab -nogui" gives (maybe more informative)

Scilab startup function detected that the function proposed to the engine is the wrong one. Usually, it comes from a linker problem in your distribution/OS.
Here, Scilab should have 'libscijvm-disable' defined but gets 'libscijvm' instead.
If you do not know what it means, please report a bug on http://bugzilla.scilab.org/. If you do, you probably know that you should change the link order in SCI/modules/Makefile.am

Is there a simple way to solve this JVM problem? (Note that the default scilab-6.0 installed in my Unbuntu has a JVM problem also and does not work. Maybe I will try to solve it another time.)

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users

_______________________________________________
users mailing list
[hidden email]
https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
-- 
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
mottelet mottelet
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Seems that emd_toolbox is not on the forge...


Le 22/12/2020 à 15:17, Stéphane Mottelet a écrit :

Hi,


Le 22/12/2020 à 14:47, Clément David a écrit :
Hello,

It looks like the scilab 5.5.2 binary is not compatible with your system (nc_putchar version error). IMHO it should be better to use the latest Scilab version and try to build dependencies by hand ; regarding the toolboxes:
 * http://forge.scilab.org/index.php/p/stftb/ is not published for 6.1 but might compile on it ; get the sources and try to build on your own machine.
 * https://atoms.scilab.org/toolboxes/emd_toolbox/ already uses the pvApiCtx symbol ; it might be possible to port it to scilab 6.1 without a big refactoring.

It is quite easy (a 10 min. job). I made the needed modifications and it compiles and loads under OSX. If you give me the rights on Gitlab I can commit the modifications this afternoon.


S.


Thanks,

Clément

De : users [hidden email] de la part de Jean-Yves Baudais [hidden email]
Envoyé : mardi 22 décembre 2020 12:36
À : Users mailing list for Scilab [hidden email]
Objet : Re: [Scilab-users] atomsLoad
 
Hi,

> You must use Scilab 5.5.2 and install the 1.2 version of the toolbox. It
> should run out of the box.

Ok. I installed it. "scilab" command and "scilab -nw" or "-nb" or "-ns" or "-noatomsautoload" give the so funny "Segmentation fault" :-( Hopefully, "scilab -nwni" works, but I only have the head_comments help and no plot. So, not so funny, but I can juggle between scilab-5.5.2 and scilab-6.1.0

"scilab -debug" does not work anymore (but I not able to use gdb!)

gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by gdb)
gdb: SCI/lib/thirdparty/libtinfo.so.5: no version information available (required by /lib/x86_64-linux-gnu/libreadline.so.7)
[...]
gdb: relocation error: /lib/x86_64-linux-gnu/libncursesw.so.5: symbol _nc_putchar version NCURSES_TINFO_5.9.20150530 not defined in file libtinfo.so.5 with link time reference

and "scilab -nogui" gives (maybe more informative)

Scilab startup function detected that the function proposed to the engine is the wrong one. Usually, it comes from a linker problem in your distribution/OS.
Here, Scilab should have 'libscijvm-disable' defined but gets 'libscijvm' instead.
If you do not know what it means, please report a bug on http://bugzilla.scilab.org/. If you do, you probably know that you should change the link order in SCI/modules/Makefile.am

Is there a simple way to solve this JVM problem? (Note that the default scilab-6.0 installed in my Unbuntu has a JVM problem also and does not work. Maybe I will try to solve it another time.)

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users

_______________________________________________
users mailing list
[hidden email]
https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
-- 
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet

_______________________________________________
users mailing list
[hidden email]
https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
-- 
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

In reply to this post by mottelet
Hi,

> It is quite easy (a 10 min. job). I made the needed modifications and it
> compiles and loads under OSX. If you give me the rights on Gitlab I can commit
> the modifications this afternoon.

Ok, but I don't how to do this, but I'm ready to learn :-) So, could you give me the step starting with "stftb" source? https://atoms.scilab.org/toolboxes/stftb/2.0.2/files/stftb-2.0.2-src.zip (I'm regular with Linux command-line, not too much with makefile)

Or, maybe there already exists a document explaining the steps?

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
Clément David-3 Clément David-3
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hello,

From the source directory, you could use `exec builder.sce` [1]. Atoms toolboxes only rely on Scilab scripts to describe and execute the build (Makefiles are used behind the scene) ; for just rebuilding the toolbox, it should be as simple as executing the builder, bumping the version and zip the directory.

Thanks,

Clément

De : users <[hidden email]> de la part de Jean-Yves Baudais <[hidden email]>
Envoyé : mardi 22 décembre 2020 16:12
À : Users mailing list for Scilab <[hidden email]>
Objet : Re: [Scilab-users] atomsLoad
 
Hi,

> It is quite easy (a 10 min. job). I made the needed modifications and it
> compiles and loads under OSX. If you give me the rights on Gitlab I can commit
> the modifications this afternoon.

Ok, but I don't how to do this, but I'm ready to learn :-) So, could you give me the step starting with "stftb" source? https://atoms.scilab.org/toolboxes/stftb/2.0.2/files/stftb-2.0.2-src.zip (I'm regular with Linux command-line, not too much with makefile)

Or, maybe there already exists a document explaining the steps?

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
mottelet mottelet
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

In reply to this post by jbaudais
Hi,

Le 22/12/2020 à 16:12, Jean-Yves Baudais a écrit :
> Hi,
>
>> It is quite easy (a 10 min. job). I made the needed modifications and it
>> compiles and loads under OSX. If you give me the rights on Gitlab I can commit
>> the modifications this afternoon.
> Ok, but I don't how to do this, but I'm ready to learn :-) So, could you give me the step starting with "stftb" source? https://antispam.utc.fr/proxy/2/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/atoms.scilab.org/toolboxes/stftb/2.0.2/files/stftb-2.0.2-src.zip (I'm regular with Linux command-line, not too much with makefile)
>
> Or, maybe there already exists a document explaining the steps?

First, uncompress the zip file. In the obtained directory, go to the
sci_gateway/c folder and remove

#include <malloc.h>

at line 8 in the file kiss_fft.h

Then in main toolbox directory, exec builder.sce as already said by
Clément. If you had no error just "exec loader.sce".

S.

>
> --
> Jean-Yves Baudais
> _______________________________________________
> users mailing list
> [hidden email]
> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users

--
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
mottelet mottelet
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

In reply to this post by jbaudais
I sent you (private message) the modified sources of emd_toolbox, for
the gateway part only. The toolbox compiles and loads, but then there
are problems due to the convention with the empty matrix that changed
since Scilab 6. I hope you will be able to fix them yourself (keep us
informed of your fixes).

S.

Le 22/12/2020 à 16:12, Jean-Yves Baudais a écrit :

> Hi,
>
>> It is quite easy (a 10 min. job). I made the needed modifications and it
>> compiles and loads under OSX. If you give me the rights on Gitlab I can commit
>> the modifications this afternoon.
> Ok, but I don't how to do this, but I'm ready to learn :-) So, could you give me the step starting with "stftb" source? https://antispam.utc.fr/proxy/2/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/atoms.scilab.org/toolboxes/stftb/2.0.2/files/stftb-2.0.2-src.zip (I'm regular with Linux command-line, not too much with makefile)
>
> Or, maybe there already exists a document explaining the steps?
>
> --
> Jean-Yves Baudais
> _______________________________________________
> users mailing list
> [hidden email]
> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users

--
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

Hi,

  Thank you for the code. I'm going to see what I can do.

--
Jean-Yves

----- Original Message -----
> From: "Stéphane Mottelet" <[hidden email]>
> To: "Users mailing list for Scilab" <[hidden email]>
> Sent: Tuesday, 22 December, 2020 17:02:48
> Subject: Re: [Scilab-users] atomsLoad

> I sent you (private message) the modified sources of emd_toolbox, for
> the gateway part only. The toolbox compiles and loads, but then there
> are problems due to the convention with the empty matrix that changed
> since Scilab 6. I hope you will be able to fix them yourself (keep us
> informed of your fixes).
>
> S.
>
> Le 22/12/2020 à 16:12, Jean-Yves Baudais a écrit :
>> Hi,
>>
>>> It is quite easy (a 10 min. job). I made the needed modifications and it
>>> compiles and loads under OSX. If you give me the rights on Gitlab I can commit
>>> the modifications this afternoon.
>> Ok, but I don't how to do this, but I'm ready to learn :-) So, could you give me
>> the step starting with "stftb" source?
>> https://antispam.utc.fr/proxy/2/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/atoms.scilab.org/toolboxes/stftb/2.0.2/files/stftb-2.0.2-src.zip
>> (I'm regular with Linux command-line, not too much with makefile)
>>
>> Or, maybe there already exists a document explaining the steps?
>>
>> --
>> Jean-Yves Baudais
>> _______________________________________________
>> users mailing list
>> [hidden email]
>> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
>
> --
> Stéphane Mottelet
> Ingénieur de recherche
> EA 4297 Transformations Intégrées de la Matière Renouvelable
> Département Génie des Procédés Industriels
> Sorbonne Universités - Université de Technologie de Compiègne
> CS 60319, 60203 Compiègne cedex
> Tel : +33(0)344234688
> http://www.utc.fr/~mottelet
>
> _______________________________________________
> users mailing list
> [hidden email]
> http://lists.scilab.org/mailman/listinfo/users

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

In reply to this post by mottelet
Hi,

I fixed the problems. The modifications are

1) emd.sci file, line 410
"//pause"
becomes
"pause"

2) emd.sci file, line 683
opt_fields = {'t','stop','display','maxiterations','fix','maxmodes','interp','fix_h','mask','ndirs','complex_version'};
becomes
opt_fields = ['t','stop','display','maxiterations','fix','maxmodes','interp','fix_h','mask','ndirs','complex_version'];

3) emd.sci file, line 794
if ~type(INTERP)==10 | ~or(convstr(INTERP)==convstr({'linear','cubic','spline'}))
becomes
if ~type(INTERP)==10 | ~or(convstr(INTERP)==convstr(['linear','cubic','spline']))

4) emd_local_peak.sci, line 32
lm = find(diff(a)~=1) + 1;
becomes
lm = find(diff(a)~=1); if lm==[], then lm=1; end

5) boundary_conditions_emd.sci file, the function moc_fliplr should be added at the end of the file
function y = moc_fliplr(x)
//  Copyright Aldo I Maalouf
if ndims(x)~=2,
disp('X must be a 2-D matrix!')
end
y = x(:,$:-1:1);
endfunction

The modification 1) is really a minor one. The remove of moc_fliplr() function in boundary_conditions_emd.sci file is very strange (this function exists in older emd_toolbox version!).

I only tested the emd() function and it gives the same output with toy examples as the ones obtained with emd_toolbox of repository 5.4 (tested on scilab-5.5.2). However, I didn't test all the paramaters of the function.

It remains one problem: the memory used always increases. If emd() function is ran multiple times, my 8G mem is fulfilled by scilab and of course scilab crashes. Maybe this comes from the modified sources of emd_toolbox done by Stephane Mottelet? (Note that there is no memory issues with emd() function of repository 5.4.)

Jean-Yves

----- Original Message -----
> From: "Stéphane Mottelet"
> To: "Users mailing list for Scilab" scilab.org>
> Sent: Tuesday, 22 December, 2020 17:02:48
> Subject: Re: [Scilab-users] atomsLoad

> I sent you (private message) the modified sources of emd_toolbox, for
> the gateway part only. The toolbox compiles and loads, but then there
> are problems due to the convention with the empty matrix that changed
> since Scilab 6. I hope you will be able to fix them yourself (keep us
> informed of your fixes).
>
> S.
>
> Le 22/12/2020 à 16:12, Jean-Yves Baudais a écrit :
>> Hi,
>>
>>> It is quite easy (a 10 min. job). I made the needed modifications and it
>>> compiles and loads under OSX. If you give me the rights on Gitlab I can commit
>>> the modifications this afternoon.
>> Ok, but I don't how to do this, but I'm ready to learn :-) So, could you give me
>> the step starting with "stftb" source?
>> https://antispam.utc.fr/proxy/2/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/atoms.scilab.org/toolboxes/stftb/2.0.2/files/stftb-2.0.2-src.zip
>> (I'm regular with Linux command-line, not too much with makefile)
>>
>> Or, maybe there already exists a document explaining the steps?
>>
>> _______________________________________________
>> users mailing list
>> [hidden email]
>> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
>
> --
> Stéphane Mottelet
> Ingénieur de recherche
> EA 4297 Transformations Intégrées de la Matière Renouvelable
> Département Génie des Procédés Industriels
> Sorbonne Universités - Université de Technologie de Compiègne
> CS 60319, 60203 Compiègne cedex
> Tel : +33(0)344234688
> http://www.utc.fr/~mottelet
>
> _______________________________________________
> users mailing list
> [hidden email]
> http://lists.scilab.org/mailman/listinfo/users
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
mottelet mottelet
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad


Le 04/01/2021 à 15:55, Jean-Yves Baudais a écrit :

> Hi,
>
> I fixed the problems. The modifications are
>
> 1) emd.sci file, line 410
> "//pause"
> becomes
> "pause"
>
> 2) emd.sci file, line 683
> opt_fields = {'t','stop','display','maxiterations','fix','maxmodes','interp','fix_h','mask','ndirs','complex_version'};
> becomes
> opt_fields = ['t','stop','display','maxiterations','fix','maxmodes','interp','fix_h','mask','ndirs','complex_version'];
>
> 3) emd.sci file, line 794
> if ~type(INTERP)==10 | ~or(convstr(INTERP)==convstr({'linear','cubic','spline'}))
> becomes
> if ~type(INTERP)==10 | ~or(convstr(INTERP)==convstr(['linear','cubic','spline']))
>
> 4) emd_local_peak.sci, line 32
> lm = find(diff(a)~=1) + 1;
> becomes
> lm = find(diff(a)~=1); if lm==[], then lm=1; end
>
> 5) boundary_conditions_emd.sci file, the function moc_fliplr should be added at the end of the file
> function y = moc_fliplr(x)
> //  Copyright Aldo I Maalouf
> if ndims(x)~=2,
> disp('X must be a 2-D matrix!')
> end
> y = x(:,$:-1:1);
> endfunction
>
> The modification 1) is really a minor one. The remove of moc_fliplr() function in boundary_conditions_emd.sci file is very strange (this function exists in older emd_toolbox version!).
>
> I only tested the emd() function and it gives the same output with toy examples as the ones obtained with emd_toolbox of repository 5.4 (tested on scilab-5.5.2). However, I didn't test all the paramaters of the function.
>
> It remains one problem: the memory used always increases. If emd() function is ran multiple times, my 8G mem is fulfilled by scilab and of course scilab crashes. Maybe this comes from the modified sources of emd_toolbox done by Stephane Mottelet?

Surely not. I just removed one obsolete #include and didn't touch
anything else. The memory leak was already present in the original source...

S.

>   (Note that there is no memory issues with emd() function of repository 5.4.)
>
> Jean-Yves
>
> ----- Original Message -----
>> From: "Stéphane Mottelet"
>> To: "Users mailing list for Scilab" scilab.org>
>> Sent: Tuesday, 22 December, 2020 17:02:48
>> Subject: Re: [Scilab-users] atomsLoad
>> I sent you (private message) the modified sources of emd_toolbox, for
>> the gateway part only. The toolbox compiles and loads, but then there
>> are problems due to the convention with the empty matrix that changed
>> since Scilab 6. I hope you will be able to fix them yourself (keep us
>> informed of your fixes).
>>
>> S.
>>
>> Le 22/12/2020 à 16:12, Jean-Yves Baudais a écrit :
>>> Hi,
>>>
>>>> It is quite easy (a 10 min. job). I made the needed modifications and it
>>>> compiles and loads under OSX. If you give me the rights on Gitlab I can commit
>>>> the modifications this afternoon.
>>> Ok, but I don't how to do this, but I'm ready to learn :-) So, could you give me
>>> the step starting with "stftb" source?
>>> https://antispam.utc.fr/proxy/2/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/antispam.utc.fr/proxy/2/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/atoms.scilab.org/toolboxes/stftb/2.0.2/files/stftb-2.0.2-src.zip
>>> (I'm regular with Linux command-line, not too much with makefile)
>>>
>>> Or, maybe there already exists a document explaining the steps?
>>>
>>> _______________________________________________
>>> users mailing list
>>> [hidden email]
>>> https://antispam.utc.fr/proxy/2/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
>> --
>> Stéphane Mottelet
>> Ingénieur de recherche
>> EA 4297 Transformations Intégrées de la Matière Renouvelable
>> Département Génie des Procédés Industriels
>> Sorbonne Universités - Université de Technologie de Compiègne
>> CS 60319, 60203 Compiègne cedex
>> Tel : +33(0)344234688
>> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/www.utc.fr/~mottelet
>>
>> _______________________________________________
>> users mailing list
>> [hidden email]
>> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
> _______________________________________________
> users mailing list
> [hidden email]
> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users

--
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

----- Original Message -----
> From: "Stéphane Mottelet" <[hidden email]>
> Le 04/01/2021 à 15:55, Jean-Yves Baudais a écrit :
>> It remains one problem: the memory used always increases. If emd() function is
>> ran multiple times, my 8G mem is fulfilled by scilab and of course scilab
>> crashes. Maybe this comes from the modified sources of emd_toolbox done by
>> Stephane Mottelet?
>
> Surely not. I just removed one obsolete #include and didn't touch
> anything else. The memory leak was already present in the original source...


I checked "diff" of sci files in emd_toolbox-1.3 and emd_toolbox-1.2.1 (the one working on scilab-5.5.2). The differences I found are about emd_ prefix added to function names (and the incomprehensible remove of moc_fliplr). I will check the c code.

Jean-Yves
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
jbaudais jbaudais
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad

The difference between emd_toolbox-1.2-1 and emd_toolbox-1.3 are the "void *pvApiCtx" added in cemdc.c, cemdc2.c, cemdc_fix.c, cemdc2_fix.c, emdc.c and emdc_fix.c. Could that be the source of the memory problem?

Jean-Yves

----- Original Message -----
> From: "Jean-Yves Baudais" <[hidden email]>
> To: "Users mailing list for Scilab" <[hidden email]>
> Sent: Monday, 4 January, 2021 16:16:42
> Subject: Re: [Scilab-users] atomsLoad

> ----- Original Message -----
>> From: "Stéphane Mottelet" <[hidden email]>
>> Le 04/01/2021 à 15:55, Jean-Yves Baudais a écrit :
>>> It remains one problem: the memory used always increases. If emd() function is
>>> ran multiple times, my 8G mem is fulfilled by scilab and of course scilab
>>> crashes. Maybe this comes from the modified sources of emd_toolbox done by
>>> Stephane Mottelet?
>>
>> Surely not. I just removed one obsolete #include and didn't touch
>> anything else. The memory leak was already present in the original source...
>
>
> I checked "diff" of sci files in emd_toolbox-1.3 and emd_toolbox-1.2.1 (the one
> working on scilab-5.5.2). The differences I found are about emd_ prefix added
> to function names (and the incomprehensible remove of moc_fliplr). I will check
> the c code.
>
> Jean-Yves
> _______________________________________________
> users mailing list
> [hidden email]
> http://lists.scilab.org/mailman/listinfo/users

--
Jean-Yves Baudais
_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
mottelet mottelet
Reply | Threaded
Open this post in threaded view
|

Re: atomsLoad


Le 04/01/2021 à 18:01, Jean-Yves Baudais a écrit :
> The difference between emd_toolbox-1.2-1 and emd_toolbox-1.3 are the "void *pvApiCtx" added in cemdc.c, cemdc2.c, cemdc_fix.c, cemdc2_fix.c, emdc.c and emdc_fix.c. Could that be the source of the memory problem?

Oh yes I didn't remember those ones. It has nothing to do with memory
allocation, this is just a required explicit argument in Scilab 6 API.
Otherwise you won't be able to compile the gateways. As I already said
the memory leak is just a feature of the original source...

S.


>
> Jean-Yves
>
> ----- Original Message -----
>> From: "Jean-Yves Baudais" <[hidden email]>
>> To: "Users mailing list for Scilab" <[hidden email]>
>> Sent: Monday, 4 January, 2021 16:16:42
>> Subject: Re: [Scilab-users] atomsLoad
>> ----- Original Message -----
>>> From: "Stéphane Mottelet" <[hidden email]>
>>> Le 04/01/2021 à 15:55, Jean-Yves Baudais a écrit :
>>>> It remains one problem: the memory used always increases. If emd() function is
>>>> ran multiple times, my 8G mem is fulfilled by scilab and of course scilab
>>>> crashes. Maybe this comes from the modified sources of emd_toolbox done by
>>>> Stephane Mottelet?
>>> Surely not. I just removed one obsolete #include and didn't touch
>>> anything else. The memory leak was already present in the original source...
>>
>> I checked "diff" of sci files in emd_toolbox-1.3 and emd_toolbox-1.2.1 (the one
>> working on scilab-5.5.2). The differences I found are about emd_ prefix added
>> to function names (and the incomprehensible remove of moc_fliplr). I will check
>> the c code.
>>
>> Jean-Yves
>> _______________________________________________
>> users mailing list
>> [hidden email]
>> https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users

--
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
12