csvread 3.0.1 exponentially slower than v 3.0.0

[David Bateman]

John W. Eaton wrote:
> On 10-Jul-2008, David Bateman wrote:
>
> | The change between the two versions was essentially to change the
> | C-string handling to C++ string handling. It also more correctly
> | duplicates the Matlab interface. This was done for this codes inclusion
> | in Octave itself.
>
> It looks like there was more than that in going from revision 4585 (a
> version attributed to Kai Habel) to 4605 (a version attributed to
> Jonathon Stickel).  In the current version, I think the problem is
> that the output matrix is resized each time a new row is encountered
> (or, if a line has more columns than are currently in the matrix).
> For the example file provided, that means approximately 180000 calls
> to resize, with most of these just adding three elements to the
> matrix.  I think we need something better than that...  I don't have
> time to work on it at the moment, but would consider a patch.
>
> jwe
>
>   
No it should only resize if the size of the output matrix changes size
as there are additional columns added, at least that was the idea..

D.


-- 
David Bateman                                David.Bateman at motorola.com
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary