Re: [Octave-bug-tracker] [bug #40341] Logical indexing into sparse matrices causes OOM

[David Bateman]

Jordi,

I'm taking this on to the maintainers list as this has nothing to do with a bug.

You seem to think I'm a false authority and you are trying to insult me with 
your comment. However, as I implemented the sparse type in Octave and probably 
90% of the sparse code is still mine the fact is I am the authority when it 
comes to the sparse code of octave.

Insulting an authority is the definition of insolence which is a childish 
behavior. 

Communities of all types survive because people are well mannered to each, so 
rather than treat you as the child you seem to want to be treated as I'll give 
a constructive response to your insult

Here are some principals for the sparse implementation

1) The row and column indices of the sparse matrix should be of the type 
octave_idx_type, limited to [0 intmax()] and not artificially limited to 
something smaller

2) numel() should have a return type of octave_idx_type and defined as the 
total number of zero and non-zero elements if the matrix and so limited to [0 
intmax()]

The corollary is that numel() can overflow for large sparse matrices. A final 
principal is that

3) If a function or operation can be implemented for large sparse matrices, 
where numel would normally overflow, then it should be. If it can't be 
implemented without numel it should fail with an intelligent error.

If you want to attack one of these principals fine we have something to talk 
about. If you think you can fix the bug cited here while respecting these 
principals and still using numel fine as well. Otherwise I see no point in 
continuing this discussion

D.
> Le 22 oct. 2013 à 02:50, Jordi Gutiérrez Hermoso <address@hidden> a écrit :
> 
> Please use the bug tracker to post updates to a bug report.  The mailing list 
> is intended as a read-only notification stream.  Info posted to this mailing 
> list address won't appear in the tracker database where it is most useful.
> 
> Update of bug #40341 (project octave):
> 
>                  Status:                    None => Confirmed              
> 
>    _______________________________________________________
> 
> Follow-up Comment #1:
> 
> Ah, this is one that you won't be able to fix by avoiding calls to numel, Mr
> Bateman...
> 
>    _______________________________________________________
> 
> Reply to this item at:
> 
>  <http://savannah.gnu.org/bugs/?40341>
> 
> _______________________________________________
>  Message sent via/by Savannah
>  http://savannah.gnu.org/
> 
>