Re: reading file

[Jonathan Stickel]

On 2/26/08 address@hidden wrote:
> From: marco cammarata <address@hidden>
> Subject: Re: reading file
> To: Michael Goffioul <address@hidden>
> Cc: address@hidden
> Message-ID: <address@hidden>
> Content-Type: text/plain; charset="us-ascii"
>
> Dear Stefan and Michael,
>
> Thanks for your suggestions.
> I found a function "textread" in the io package of octave-forge that do
> the job reasonably.
>
> I then created an octave "library" to read the file.
> After having loaded the library (source "lib.octave")
> then
> data=read_diagnostic(0);
> would do the job returning a structure.
> On my computer Xeon(TM) CPU 3.06GHz with octave 3.0
> it takes 0.5 second to read the attached diagnostics.log
> What is funny is that if i skip the first 150 lines (out of 160)
> with the following syntax
> data=read_diagnostic(150);
> It takes the same time ! (note that the second execution of the command
> is always a bit faster, I'm referring to the first call).
>
> Do you think a oct-file might speed up even further ?
> I have no idea on how to write oct file,
> any suggestion of something to read ?
>
> regards,
> marco

From what I recall from your original message, I think you are looking 
for a function like Matlab's "importdata" that can read the columns of a 
data file into a structure with fieldnames given by the header of the 
file.  I have plans to write exactly that function as a c++/oct file 
(for the reason of speed), but I am not sure when I will have the time. 
  My plan is to adapt the the code from dlmread.cc that was recently 
added to octave-forge.  You are welcome to look at that code if you are 
interested in trying to make some progress yourself.

Regards,
Jonathan