Re: misbehavior in Octave 2.9.4 for OS/2

[Benedict G. Archer]

Also a long-time user of octave and eCS/OS2, but haven't have the last OS2 
octave binaries 2.15 (?) installed for awhile because I've had access to Matlab 
on a work system.  Had started efforts to compile octave for OS2 a number of 
times but never made much headway--the project was more than the simple 
compiles I knew how to do.  Installed Dmitry Froloff's build as soon as I 
learned of it (from Steven Levine).  Already had gnuplot so tried to set things 
so octave would use the existing gnuplot rather than what it installs.  Result 
at the moment, only the built-in octave functions work--sin(x)/cos(x) work; 
sec(x) isn't found--same for fft, eig, svd, rand, plot and so on.  I can get 
things working, but I'd like to know what a correct or recommended installation 
looks like.  My intent is to find and study the REXX install script that came 
with Klaus Gebhardt's earlier octave builds for OS2 to see how to set 
environmental variables, what should be in .octaverc, octaverc, termcap!
 , which of these are needed etc.  Or, is there a better place to look for 
this?  

Just installed linux (SuSE 10.1) with main motivation being to see how an 
octave installation looks in linux, but still finding my way around this new, 
to me, OS.

Thanks,
Ben Archer


In <address@hidden>, on 06/16/2006 
   at 09:51 AM, "Steven Levine" <address@hidden> said:


>In <address@hidden>, on 06/16/06
>   at 08:37 AM, address@hidden (Dushan Mitrovich) said:

>Hi,

>As both an eCS/OS2 user and sometime Octave user, perhaps I can help get
>you started.

>>Yesterday I downloaded version 2.9.4 compiled for OS/2 by Dmitry Froloff
>><http://froloff.homeip.net/zips/octave-2.9.4-bin.zip>, and installed it.
>>My OS is eComStation 1.2, the current version of OS/2.

>First, be aware that Dmitry's build is very new and very lightly tested. 
>The good news is now that eCS/OS2 has a capable gcc build chain, the
>patches are minimal and unobtrusive.  Over time they might be integrated
>into the Octave sources.

>It's on my list to evaluate the build in more detail, but to date, all
>I've had time for is to review the patches.

>>   When defined from the command line for a vector 'a = 0:pi/10:2*pi' it

>Just to clarify the terminology, defining a function is not the same as
>invoking a function.  The former just records the function defintion for
>later use.  Something like

> a = mx + b

>is just an expression.

>  a = f (x)

>is an expression that invokes a function.

>>   Is this normal, or a consequence of my function definition?

>It is normal, in the sense that expression results always display unless
>suppressed by terminating the expression with a semi-colon.

>>   error: no such file,
>>'I:/octave/I:\octave/scripts/deprecated/is_vector.m'

>Octave is from the unix world.  You need to use forward slashes in your
>settings.  I suspect you need to update your LOADPATH setting in
>octave.ini.

>>   and discovered that asking for 'sinc(1)' kept producing the exact same
>>   message.  Only after I quit Octave, then restarted it, did it recog-
>>   nize the change and produce a different message.

>This is either a problem in Dmitry's port or with your EDITOR setting. 
>What is supposed to happen is that Octave is supposed to notice the
>timestamp change and recompile the .m file.  Since E is a PM application,
>octave may have considered the edits complete before you were actually
>done.  Try setting EDITOR to tedit.exe or some other VIO mode editor.  If
>the behavior persists, I would discuss the issue with Dmitry.

>As a workaround you can tell Octave to forget the function with clear.

>>   between Octave's path references using a forward slash '/' and OS/2's
>>   using a backward slash '\', but I'm not sure how to resolve that.

>You need to use forward slashes in your configuration settings.

>>        SYS0003: The system cannot find the path specified
>>        < a one-line definition >

>This may be related to your settings or it may be a problem with the
>port.  SYS0003 means path not found.

>>        warning: help: Texinfo formatting filter exiting abnormally
>>        warning: help: raw Texinfo source of help text follows...

>>   except that for 'help edit_history' there are an additional two lines:

>>        WARNING: terminal is not fully functional:
>>        -   (press RETURN)

>You need to correct your TERMCAP/TERMINFO settings.

>>   This warning also appears when I ask for 'history 100'.

>What happens inside Octave when you ask for help is Octave looks for
>documentation in one of several places.  The documentation is written in
>texi format.  This is run through texi2info (sic) to generate info
>formatted data and info is invoked to display the formatted data.

>What happens when you invoke

>  help -i history

>Does the info formatted documentation display properly?

>Regards,

>Steven



-- 
-----------------------------------------------------------
Benedict G. Archer<address@hidden>
-----------------------------------------------------------