Re: crash testing luinc.cc

[Jaroslav Hajek]

On Thu, Sep 30, 2010 at 7:31 PM, John Swensen <address@hidden> wrote:
>
> On Sep 30, 2010, at 3:18 AM, Jaroslav Hajek wrote:
>
>> On Wed, Sep 29, 2010 at 10:23 PM, John W. Eaton <address@hidden> wrote:
>>> On 29-Sep-2010, Jaroslav Hajek wrote:
>>>
>>> | On Wed, Sep 29, 2010 at 6:58 PM, John W. Eaton <address@hidden> wrote:
>>> | > On 29-Sep-2010, bpabbott wrote:
>>> | >
>>> | > | My tip is
>>> | > |
>>> | > | changeset:   11060:b099acf06b55
>>> | > | tag:         tip
>>> | > | user:        John W. Eaton <address@hidden>
>>> | > | date:        Wed Sep 29 04:25:57 2010 -0400
>>> | > | summary:     make_int.cc: don't pass arg to mlock
>>> | > |
>>> | > | When testing luinc.cc, I get a bus-error on MacOS X.
>>> | >
>>> | > I checked in the following changeset which I think fixes the problem:
>>> | >
>>> | >  http://hg.savannah.gnu.org/hgweb/octave/rev/c33b7054f1f9
>>> | >
>>> |
>>> | In fact it was just off-by-1 error:
>>> | http://hg.savannah.gnu.org/hgweb/octave/rev/b721e12140cc
>>>
>>> Thanks for fixing that.  I didn't investigate, other than to notice
>>> that I had actually intended to use the getfield function but forgot
>>> to make the change in a few spots.
>>>
>>> I plan to work my way through the rest of the conversion to
>>> octave_scalar_map and octave_map in the next few days.
>>>
>>
>> Great. I replaced several occurrences of Octave_map today, to save you
>> some work. Important left-overs include the graphics stuff and
>> ls-mat5. Maybe it would be possible to mark Octave_map as obsolete
>> already in 3.4. It also seems to turn out that at least 50% of
>> Octave_map uses are suitable for the simpler scalar interface of
>> octave_scalar_map (which was what I hoped for from the start).
>>
>> regards
>>
>> --
>> RNDr. Jaroslav Hajek, PhD
>> computing expert & GNU Octave developer
>> Aeronautical Research and Test Institute (VZLU)
>> Prague, Czech Republic
>> url: www.highegg.matfyz.cz
>>
>
> So, should I avoid using Octave_map and instead use octave_map in my work 
> with imread/imwrite?  The current code uses an Octave_map to store the 
> parameter name/value pairs.  I saw in the oct-map.h file that it says the two 
> are interchangeable, but is one now preferred over the other?
>

If your structs are scalar, use octave_scalar_map. Only if you truly
need a struct array, use octave_map.
Octave_map will still work, but will eventually be deprecated.

hth

-- 
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz