On Fri, 17 Nov 2006, Tim Jenness wrote:
>> The changes look to be pretty minor. Additionally, I don't actually
>> use much of this package apart from the header handling parts - the
>> bulk data handling code in nom.tam.* out to be not efficient enough for
>> what I wanted to do, so I now use custom code to do most of the
>
> *cough*. So I take it that making tamfits work better wasn't possible?
Not without considerable redesign of the interface. nom.tam.fits is at a
higher level than what I need, and is not amenable to the kind of lazy
I/O required for scalability to very large data sets. Hope the cough
clears up.
>> things that package does. I've tried running my unit tests against the
>> 0.99.1 [tam]fits.jar file from Tom's archive and they seem to go OK.
>
> Does this mean you aren't even going to send him your patch?
It's not a particularly clean patch (because it's not that clear what
the code it's patching is trying to do) and he may prefer the existing
behaviour, so I wasn't planning to. Also, I'd talked to Tom some
years ago about submitting code for inclusion in this package and
while he didn't actually say no, he didn't sound very enthusiastic.
However if you think I should I will if/when we decide whether we're
going to upgrade our tamfits to the latest release.
Mark
--
Mark Taylor Astronomical Programmer Physics, Bristol University, UK
[log in to unmask] +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/
|