[SciPy-dev] Three cheers for Travis!

[Travis Oliphant]

Chris Fonnesbeck wrote:

>On 11/30/05, Ed Schofield <schofield at ftw.at> wrote:
>  
>
>>I'd like to thank Travis for his tireless work answering feature
>>requests, fixing bugs, and generally doing a superb job.  Hip hip
>>hooray!  SciPy is becoming very good, very fast.
>>
>>    
>>
>
>I will second this motion. We all appreciate your work, Travis.
>
>  
>
*Blush*

Thank you for the kudos. I do appreciate them.

I'm spending a lot of time right now getting scipy_core to the point 
that it can be adopted without hesitation by lots of people.   At the 
start of the year, I won't have this kind of time, so I'm hoping that we 
can get as many wrinkles ironed out as possible.  

There are still some things pending.  I just added support for character 
arrays that needs some testing.  The ndchararray is a subclass of the 
ndarray that soups up the string and unicode support (it allows 
comparison functions element by element and adapts all the methods of 
strings and unicode objects to work element-by-element).

It makes use of the broadcast iterator that has been exposed to Python 
so that broadcasting is supported rather like ufuncs.  I'm really 
pleased with some of the abstractions that we've been able to extract 
from old Numeric and make more universally accessible.   I'm anxious for 
people to start to use them and see if they work as expected and/or can 
be made better.

I'm working on the records.py module right now and adding a few things 
to make that a bit easier. 

Other tasks still on the back-burner:

  * improving coercion of strings to other types
  * giving array scalars their own optimized math
     with supported error-control flags instead of going through
     the slower ufunc machinery

-Travis