[Numpy-discussion] Ironclad 0.8 released

[William Reade]

Hi Tim

Thank you for your interest! To answer your various questions order  
of approximately increasing complexity:

Most importantly: the currently available version of Resolver One  
does not have numpy support built in, and for various reasons I  
wouldn't advise trying to integrate Ironclad with v1.3; please wait  
for 1.4 :-).

I just downloaded R and rpy2; they don't work with Ironclad at the  
moment, and I'm not yet sure why.

Ironclad shouldn't require any special installation: just copy the  
ironclad package (containing __init__.py, python25.dll and  
ironclad.dll) to wherever you normally put packages for IronPython (I  
personally have my IRONPYTHONPATH pointing to the appropriate  
subdirectories of a standard Python 2.5 install). Clearly my  
documentation is currently inadequate; I'll have to work on that for  
the next release.

RO uses .NET DateTime objects by default, but it's perfectly possible  
to use python datetime objects instead; I can confirm that you can  
put both those datatypes into numpy arrays.

While RO doesn't have an explicit CSV/TSV import option, you should  
be able to paste data in those formats directly into your worksheets;  
and, as soon as we have numpy integration out in the wild, you should  
be able to create ndarrays directly from your data files.

Scipy support is coming along surprisingly nicely; I'd sort of  
forgotten about it until a couple of people asked me about it earlier  
today, and it turns out -- to my surprise and delight -- that it took  
very little work to get the SVN head running over 800 of the scipy  
tests (even if you can't just 'import scipy' without errors). Sadly,  
a lot of the modules seem to use scipy.sparse, which I can't  
import... yet ;-).

So, we haven't built any scipy support into Resolver One yet, but  
hopefully we'll be able to do so without too much pain.

Finally, I'm glad you like my screencast :-). I hope I haven't missed  
anything; let me know if I have, or if I can help with anything else.

William

On 29 Jan 2009, at 23:46, Tim Michelsen wrote:

> Hello William,
>
>> * A few performance improvements.
>> * Over 900 NumPy tests now pass: in fact, almost all the tests  
>> from the core, fft, lib, linalg, ma, oldnumeric and random  
>> subpackages.
> I have some questions here for the science interested. I hope that  
> they are not too specific:
>
> * Can you also use scipy with IronClad/ResolverOne (RO)?
> * Can you recognize arrays with datetime objects in RO?
> * How is datetime information handled in RO?
> * What about rpy?
>
> If all the above were working in RO, it would be terrific!
>
>> Ironclad grows more stable and mature with every release, and I  
>> urge IronPython users to try it out and share their impressions:  
>> feedback, whether positive or negative, is always welcomed.
> You may find some information here:
> http://groups.google.com/group/python-excel/msg/eb349c229696c7f7
>
> I would also recommend this nice & straightforward video
> http://www.resolversystems.com/screencasts/numpy/
>
> Kind regards,
> Timmie