Hello

> (Or is it the oct2mat package that was loaded accidentally - you did 
> write that you explicitly didn't install it?)


Installer of octave-3.2.4/mingw32 makes oct2mat autoload if all octave-forge 
packages are installed.

This is known as the oct2mat issue of the current octave for windows package.

Please see

http://wiki.octave.org/wiki.pl?OctaveForWindows
1. Standalone ports
 1.1. Octave 3.2.4 for Windows MinGW32
  Additional important topics found after the release:

Regards

Tatsuro

--- Philip Nienhuis  wrote:

> <I forgot to include the list...>
> 
> 
> Hi there Ramon,
> 
> sorry for long reply but it's a reply to 2 mails in one :-)
> 
> Ramon Bueno wrote:
>  > Hello Phillip,
> 
> ... with one "l" please...  :-)   (Don't worry, just nitty gritty)
> 
>  > a) Down below are the additional COM/ActiveX & java tests you requested
>  > and their results.
> 
> Actually all test results look good.... read on.
> 
>  > b) Regarding the script I left running last night: This morning I saw
>  > these final messages in the Octave window. (plus, when I "quit" Octave I
>  > saw a flood of messages fly by, very fast, could not read them). In
>  > addition, the output Excel file that last night was being filled with
>  > separate worksheets by the Matlab code (I recall seeing 5 sheets with
>  > data in them)....well today the file is still there, but it is 0 KB in
>  > size and is basically empty. It seems as if perhaps the closing code
>  > wiped it out?
> 
> No, I strongly suspect something in your ML scripts, see below.
> 
> But your previous batch of error messages shows that you'd better 
> increase the Java virtual memory settings, see below too.
> 
> <snip>
>  > octave.exe:4> exl = actxserver ('Excel.Application')
>  > error: `actxserver' undefined near line 4 column 7
> 
> This means the windows package hasn't been loaded or installed.
> 
>  > octave.exe:4> *pkg load java*
> 
> No errors, so this is good.
> 
> BTW why are there a leading and trailing asterisk?
> 
>  > octave.exe:5> *pkg load windows*
> 
> No errors, this is good.
> 
>  > octave.exe:6> *exl = actxserver ('Excel.Application')*
>  > exl =
>  > <COM object _Application (0x0x89887c)>
> 
> This is good.
> 
>  > octave.exe:7> *delete ('exl')*
>  > warning: delete: no such file: exl
> 
> exl = a "variable" (COM object), not a file name, so shouldn't be 
> enclosed in quotes.
> I forgot to mention that before delete () one should enter "exl.Quit()" 
> (w/o quotes), sorry for that.
> 
>  > octave.exe:8> *xls1 = xlsopen ('TEST_MATLAB_input.xls', 0, 'com')*
>  > Excel/COM interface requested... Excel (COM) OK.
>  > xls1 =
> <snip>
> 
> This is good. So COM/ActiveX & invoking Excel works.
> 
> <snip>
>  > octave.exe:11> *tmp = java_new ('java.lang.String', 'abc')*
>  > tmp = abc
> 
> Good, your java works.
> 
>  > octave.exe:12>* javaclasspath*
>  > warning: split is obsolete and will be removed from a future version of
>  > Octave;
>  > please use strsplit instead
> 
> Known "buglet" of the java package, doesn't harm.
> 
>  > C:\Octave\3.2.4_gcc-4.4.0\bin\dom4j-1.6.1.jar
> <...etc...>
> <snip>
> 
> Perfect.
> 
>  > octave.exe:13> *xls2 = xlsopen ('TEST_MATLAB_input.xls', 0, 'poi')*
>  > Java/Apache POI interface requested... Java/Apache (POI) OK. (& OOXML OK)
>  > xls2 =
> <...etc...>
> <snip>
> 
> Perfect. The java POI interface for Excel works OK.
> 
>  > octave.exe:15*> xls3 = xlsopen ('TEST_MATLAB_input.xls', 0, 'jxl')*
>  > Java/JExcelAPI interface requested... Java/JExcelAPI (JXL) OK.
>  > xls3 =
> <...etc...>
>  > octave.exe:16> *xlsclose (xls3)*
>  >
>  > ans = [](0x0)
> 
> Perfect. JExcelAPI, the other java-based Excel interface, also works.
> 
> 
> My conclusion is that there's nothing wrong with your setup, except that 
> the windows & java package should be loaded automatically (see below for 
> hints).
> This means the problem probably is in your Matlab scripts. I'm afraid 
> you will have to clean up & debug your ML-scripts first.
> 
> As a first shot I had a look at the list of error messages you sent me 
> earlier on:
> 
> START============================================================
>  > warning: single quote delimited string near line 396 of file 
> C:\Octave\3.2.4_gcc
>  > -4.4.0\share\octave\packages\io-1.0.11\oct2xls.m
> 
> I think this implies that in your Matlab scripts there's an error with 
> properly enclosing names in quotes. Single or double quote shouldn't 
> matter; Matlab only knows about single quotes, Octave does both, but " 
> and ' cannot be mixed around one and the same string.
> 
>  >
>  > warning: potential Matlab compatibility problem: " used as string
>  > delimiter near
>  > line 399 offile C:\Octave\3.2.4_gcc-4.4.0\share\octave\packages
>  > \io-1.0.11\oct2x
>  > ls.m
> 
> ...Very intruiging....
> Some script is monitoring the octave scripts for ML compatibility. That 
> is Not Good, not even necessary, and probably even harmful.
> I'd check all the ML scripts and shut off this ML compatibility checking 
> interference.
> 
> (Or is it the oct2mat package that was loaded accidentally - you did 
> write that you explicitly didn't install it?)
> 
>  >
>  > warning: implicit conversion from string to real N-d array
> 
> ...I suppose a consequence of the same issue.
> 
>  >
>  > ncols =  1
> 
> ...Yep, a stray harmless echo that has been fixed (in svn) several 
> months ago.
> But by coincidence it IS informative in your situation - *do* you or 
> your scripts use/specify spreadsheet ranges of only 1 column wide? 
> (would be a nice check)
> 
>  >
>  > warning: concatenation of different character string types may have
> unintended c
>  > onsequences
> 
> ...Consequence of same issue as above I think.
> 
> ============================================================END
> Well, I'll have to draw the line here. Debugging scripts originating 
> from ML is not quite my responsibility as IO package maintainer  :-) :-)
> Nevertheless your reports are very useful to me.
> Please remove the ML compatibility checks first and try again.
> 
> OK, from your earlier mail:
> 
>  >     <snip>                             I then tried "pkg load all"
>  > in a new session and Octave similarly stops working but no messages
>  > appear.
> 
> IIRC the ga, octcdf & ?communications? packages interfere with the 
> Windows package or had other fatal issues. See the README of the 
> octave-3.2.4 MingW binary package you downloaded & installed.
> 
> Please restart octave and using "pkg list", make sure that these 
> packages + oct2mat package are NOT loaded (have no asterisk mark). Use 
> "pkg unload <pkgname>" one by one if needed.
> 
> Then, to make sure that java pkg is loaded automatically:
> pkg rebuild -auto java  (will settle it once and for all)
> 
> When that's done, make a plain text file called "java.opts" (w/o quotes) 
> and put it in <octave>\share\octave\packages\java-<version>, containing 
> just these 2 lines:
> 
> -Xms16m
> -Xmx512m
> 
> and restart octave. This will increase java's working memory 
> availability from 16 to 508 MiB. If you have fairly big spreadsheets 
> this is a fairly vital setting.
> Don't worry, this amount of memory is only "pre-reserved", not 
> immediately grabbed, and only piecewise borrowed from and returned to 
> the windows (and other OS's) memory managers by java.
> 
> To make sure that windows is loaded automatically, enter the pkg load 
> command in octaverc as I mentioned in an earlier mail.
> Do not do it like the java package; my suggestion to put it in octaverc 
> is actually a workaround for a bug affecting the octave-3.2.4 windows 
> binary.
> 
>  >
>  > - Java jre6 is in both \Program Files and\Program Files (x86). I'm
>  > not sure how jre6 relates to the 1.6 version you mention.
> 
> jre6 = Java 1.6.x.x.
> It's all OK.
> 
> Hopefully you'll manage to fix your ML scripts. There's something alien 
> in there that is quite unfriendly to non-ML scripts. I'm curious as to 
> what it actually is.
> 
> If you don't mind, can you please strip the various xlsread / xlswrite 
> calls from your scripts and mail them to me, so I can check if those 
> calls match octave's calls (which /should/ be ML-compatible)?
> 
> Best wishes & thanks,
> 
> Philip
> 
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by 
> 
> Make an app they can't live without
> Enter the BlackBerry Developer Challenge
> http://p.sf.net/sfu/RIM-dev2dev 
> _______________________________________________
> Octave-dev mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/octave-dev
> 


--------------------------------------
Get the new Internet Explorer 8 optimized for Yahoo! JAPAN
http://pr.mail.yahoo.co.jp/ie8/

------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev 
_______________________________________________
Octave-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/octave-dev

Reply via email to