[Rock-dev] Statistics class in base

[Jakob Schwendner]

On 05/14/2012 11:27 AM, Alexander Duda wrote:
> On 05/14/2012 11:13 AM, Jakob Schwendner wrote:
>> On 05/14/2012 10:58 AM, Alexander Duda wrote:
>>> On 05/14/2012 10:37 AM, Jakob Schwendner wrote:
>>>> On 05/14/2012 10:31 AM, Alexander Duda wrote:
>>>>> On 05/14/2012 10:26 AM, Jakob Schwendner wrote:
>>>>>> I would rather use a more generic package for now, since it is really
>>>>>> just a single class that doesn't merit a full package. Maybe
>>>>>> tools/numeric (including alexes stuff?)
>>>>>>
>>>>>> Jakob
>>>>> There is already a folder for image_processing. Maybe we can put
>>>>> numeric_tools into this folder for now and rename the folder
>>>>> image_processing to processing in the long run.
>>>> I would see numeric_tools as something that could be used by
>>>> image_processing, but would not be limited to it (e.g. a lot of the
>>>> packages in slam could benifit). Changing image_processing to processing
>>>> I am also not a fan of. What's the problem with base/numeric or
>>>> base/numeric_tools?
>>> Base is a very generic category therefore if there is any other fitting
>>> to the proposed package I would go for the latter one. Other than that
>>> there is nothing wrong with base/numeric(_tools).
>> I would argue it is as generic as a spline library (which is even in
>> base/types).
>> Unless there are objections maybe you could move your package to
>> base/numeric? (which btw, we could also move the sisl stuff to at some
>> point).
> Ok (moving the stuff to base/numeric)
I've created the base/numeric package now. It's a good place to add 
algorithms, that are generic and don't have any dependencies other than 
base/types. I'll have a couple of candidates that could go there that I 
will move over the time. Please make sure not to make this a collection 
for stuff that you don't know where else to put, and maintain code 
quality (e.g. tests whenever possible a.s.o)

cheers,

Jakob