On Thu, 24 Nov 2005, Tim Jenness wrote:
>>> 1. I can't test it since I don't have access to IDL. Can someone
>>> give it a try?
>>
>> Attempting that, but it hasn't worked for me yet (core dumps in NDF).
>> I'm doing a full tree rebuild as the error looks silly on closer
>> inspection. Report back tomorrow.
>>
>
> Sorry about that. I saw your patches. Not sure how I missed some of those.
Still struggling with this one. For some reason the names of the HDS files
are getting lost en-route from IDL.
>> If you're planning to add some support for ragged arrays (padded to the
>> maximum length) then I guess char** is right, or better still *char[]
>> (pointer to array of pointers), but otherwise I'd say it looks better as
>> simple char* for these, potentially, multidimensional arrays.
>>
>
> Yes. I'm edging towards keeping datGetC (multi-dim) as it is, nul-terminating
> datGet0C and (and this is the controverial bit) make datGet1C return a C
> *char[]. The complication is the memory management.
> There is no C datGet1C yet so there is no history to deal with.
Does make sense from C, I expect all those multidimensional arrays of
non-NULL terminated strings will be the actual puzzler.
>> I've committed the GAIA and SPLAT fixes. BTW, SPLAT used ndfXloc too!
>> JNIHDS looks a longer job (is Mark volunteering?).
>>
>
> Thanks. I fixed sextractor. ICL is "done" but I don't know enough ICL to know
> whether I am exercising the code or not...
I've exercised this with PONGO and uncovered a few problems. Committed the
fixes.
Cheers,
Peter.
|