Re: Octave 2.1.64 broken?

[David Bateman]

Vadim Gutnik wrote:

> I submitted a nearly identical bug for builds on two different
> architectures of 2.1.64. (January 11 and 15, "isascii problem").
>
> There has been no response.
>
> Is 2.1.64 known to be broken? Is there a fix or workaround or stable
> 2.1.x version to build from source? The Debian 2.1.35 build works, but
> I don't see those sources anywhere convenient on the octave site.
>
>        Vadim
>
>
>  
Patience...... Why exactly is the fact that isascii(1) doesn't work a 
bug? This function is supposed to identify the elements of a character 
matrix that are ascii (ie. between 0 and 127). So the fact that it 
doesn't take real args is a missing feature if it is anything...

In any case, I believe the following patch address your problem, whether 
or not it is acceptted depends on whether John interprets your problem 
as a bug or not...

D.

--
David Bateman                                address@hidden
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

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

*** src/ov-mapper.cc~   2004-12-28 04:36:14.000000000 +0100
--- src/ov-mapper.cc    2005-01-27 13:59:52.065816820 +0100
***************
*** 112,118 ****
    // XXX FIXME XXX -- is_real_type can return true.  Should it really
    // work that way?
  
!   if (arg.is_real_type () && ! (arg.is_string () && ch_map_fcn))
      {
        if (arg.is_scalar_type ())
        {
--- 112,119 ----
    // XXX FIXME XXX -- is_real_type can return true.  Should it really
    // work that way?
  
!   if (arg.is_real_type () && (d_d_map_fcn || d_b_map_fcn) && 
!       ! (arg.is_string () && ch_map_fcn))
      {
        if (arg.is_scalar_type ())
        {