Re: csv2cell

[Philip Nienhuis]

Hi Houtan,

1. Please do not email developers & contributors privately; use 
themailing lists.

Read on....

Houtan Bastani wrote:
> Hello,
>
> We make use of your csv2cell function in our program, dynare.
>
> I hope I’m not bothering you, but I was curious as to why you limit the number 
> of characters per line in the .csv file to be 4096. A priori, this seems like 
> an unnecessary constraint. But, of course, you wouldn’t have made that 
> constraint if it weren’t necessary, so I was just wondering why it’s there.

I didn't limit that line length; the io package (of which I'm the 
maintainer) "inherited" the csv2cell & cell2csv functions from another 
octave-forge package and since then I have debugged them and added some 
more functionality.
So no, I have no reason for the line length limit other than that I 
never saw any reason to touch it.

That said, I can imagine that text files with lines longer than 4096 
characters have limited use for humans (reading). But possibly they are 
useful for communication between programs w/o human interference.

FYI, the textread / textscan functions also have a line length limit of 
4096 chars (imposed by some EOL testing routine).

If you require longer line lengths for csv2cell, it is very easy to fix 
that yourself:
In the source code of csv2cell, there's a constant MAXSTRINGLENGTH, 
initialized to 4096.
Just change that number to anything you think is reasonable, also fix 
the error message a little lower (that currently claims max line length 
is 4092 - oopsie a buglet) and rebuild (in Octave):

mkoctfile csv2cell.cc

and put the csv2cell.oct file into the io package's bindir (do
which csv2cell   ## again in Octave
to find out where it is).

It may be that you hit system limits for too long line lengths; if so, 
please inform me of that.

Good luck,

Philip