Mon 23 Dec 2013 10:30:53 PM UTC, comment #1:
First off, if you want to read simple file formats like .csv, just use the proper tools for it like csvread and/or dlmread. That's a lot faster anyway than invoking overly "heavy" functions as xlsread.
Or even better, use csv2cell (also in the io package), it's also a lot faster and it can even read heterogeneous .csv files (mixed text and numeric data).
If you run Octave on Windows with MS-Excel (or in any OS with Java & LibreOffice / OpenOffice.org) installed, xlsread can read any file format that is supported by Excel (LibreOffice / OOo, resp.), including .csv.
What Octave's xlsread doesn't do is fall back to csvread if no Excel (or LO/OOo) was found, indeed. I've been very reluctant to mimic this sort of Matlab code bloat in Octave's spreadsheet functions for various reasons.
BTW 1: I noted you do have the UNO interface supported (LibreOffice/OOo). It should be possible to run xlsread like:
[num, txt, raw, lims] = xlsread (<filename.csv>, 1, "", "UNO")
(and it'll give you the desired overhead ;-) )
Please check if this works OK for you.
I think (but I'm unsure) that this should have happened automatically, but failed because of this:
BTW 2: That POI interface message is a bug / red herring, please ignore. It has been fixed in io-1.2.4 IIRC
---------------
As there are very suitable alternatives/workarounds I'll provisionally lower priority/severity to "wish"; adapted title
|