commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Phil Steitz <p...@steitz.com>
Subject Re: [all] Math needs a "user" email list.
Date Fri, 13 Aug 2004 22:57:52 GMT
Stephen Colebourne wrote:
> I would suggest that your argument is more for Apache Math than for a new
> mailing list. (Note Apache Math, not Jakarta Math)
> 
> For me, [math] goes beyond the role of a simple library of common code.

Not that far, IMHO.  Most of what is there now is basic, widely useful 
utility code.

> Instead it presents a wealth of detailed mathematical code. The user of the
> library needs some mathematical knowledge to get the best from the library.

Not very much. Default impls, etc. are designed to make it simple to do 
simple things.  If this is not true, we have not done a good job designing 
and documenting the code.

> Related to this is that the developers and advisors will be drawn from a
> different (less java/programmer centric position).

I respectfully disagree here.  We do need some targeted help with 
numerics, but what we really need is Java developers with math proficiency 
and need (for math functionality).

> 
> Secondly, there is no reason why [math] could not be implemented in C, C# or
> another language as far as I can see. (There is no general rule here, [lang]
> and [collections] are java focussed, but [codec] could be multi-language.

I disagree.  I see [math] as very Java-focussed.  The whole point of 
[math] is to provide ASF-licensed pure Java implementations of basic math 
functionality.

> 
> One option that could be considered in to use the package name
> org.apache.math for your 1.0 release even if you are still hosted in
> commons. It avoids problems later.

I would prefer to keep commons in the package name.  Later, if we toss out 
the current PROOPOSAL 
(http://jakarta.apache.org/commons/math/proposal.html) and try to become 
an "Apache math library" a new (multi-language) project could be started 
in the incubator, using some commons-math components for Java 
implementations.  Even then, I would see commons-math as a valuable 
commons component.
> 
> So, consider me -0 to a commons-math-user at the moment.
> 
> Stephen
> 
> ----- Original Message -----
> From: "Mark R. Diggory" <mdiggory@latte.harvard.edu>
> 
>>Do you get my argument though?

I get the argument, but I disagree with it. As Brent points out below, 
there are plenty of math forums out there for general discussion of 
algorithms and other math topics. I see no reason for Apache to host yet 
another one of these. I don't know what you mean by "non-developer" users, 
since our users are by definition Java developers. Discussion here should 
focus on issues related to the [math] code base.  From time to time, this 
discussion will naturally include some math content.  Since we are trying 
to stick to established algorithms and basic functionality, I would 
actually see it as a red flag if we ventured off into extended 
mathematical discussions.

For the reasons that I mentioned above, I do not want [math] to be split 
out from j-c. I will cut a RC for 1.0 this weekend and I would prefer that 
we focus on getting the release out rather than how the user list works at 
this time.

Phil

>>
>>I'm suggesting that creating the new list will help promote maths
>>'non-developer' user base who are interested in algorithms and stability
>>(not because the load is too high). Individuals who are not interested
>>in usage of other commons components per se. Individuals like Kim or
>>David (or anyone from the jama or colt lists), who may not necessarily
>>be developing Apache code, but could supply excellent discussion on the
>>subject area. It would really help to promote the project and attract a
>>many individuals who are knowledgeable in the field of mathematics, thus
>>offering expert feedback to help improve the product. A math library is
>>a much different beast than a collections api, discovery or bean
>>utilities; certain standards are expected. I want a list where such
>>things can be discussed without risking being bothersome to the commons
>>developer/user lists with overly abstract discussion on mathematics and
>>not overwhelming such math users with discussion from other commons
>>sub-groups.
>>
>>I think providing a user friendly environment is important here, and
>>someone barking "use a filter stupid, its not our fault" is really not
>>very user friendly, ASF has the facilities and can do better.
>>
>>-Mark
>>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-dev-help@jakarta.apache.org


Mime
View raw message