commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gary Gregory <garydgreg...@gmail.com>
Subject Re: [ALL] The Commons Math issue
Date Tue, 18 Apr 2017 00:54:39 GMT
On Apr 17, 2017 3:32 PM, "Gilles" <gilles@harfang.homelinux.org> wrote:

Hi Ralph.


On Sun, 16 Apr 2017 10:55:49 -0700, Ralph Goers wrote:

> Gilles,
>
> What is your vision on where things should end up?  Can you identify
> what new commons sub-projects we will have?  If it is just 3 or 4 I
> have no problem with that.
>

We have
 * RNG

We will have
 * Numbers

We might have
 * SigProc
 * Clustering


But if we are going to have 10 sub-projects
> then I really feel like it should be done as:
>
> A:  Commons Math with Commons Math RNG, Commons Math Numbers, Commons
> Math XXX, etc.
>

Do you mean a modularized "Commons Math"?
If so, my opinion is that it is not feasible at this point.


B: Math TLP with RNG, Numbers, XXX, etc.
>

Also not feasible at this point:
1. There are too few developers to support all of the code in
   "Commons Math", and I've the feeling that there won't be enough
   volunteers to help this TLP. [cf. James Carman's offer that had
   been turned down by the PMC.]
2. The amount of code that is supported by the current team is so
   small that "Commons" is the perfect home for these few new
   components.


I have a few reasons for feeling this way:
>         1. Although you seem to disagree, these components do seem to be
> related in that they all have something to do with Mathematical
> concepts.
>

It's hard to deny that, but the same goes for other components
that don't seem to trouble anyone (RDF, Compress, Crypto, Functor).


        2. It is easier for users to find if they are all grouped together.
>

By the same token, why not group all of "Commons" in a single
maven project?


I want to do that to create an Uber jar and site with Javadocs.

Gary

Doesn't make sense?  Then it also does not make
sense to group things under the sole "math" category: it's just
too broad and does not help users discover whether there is
something fit for them.



> That said, all of this is just housekeeping and can be addressed at
> almost any time.
>

Indeed.
As Ray also pointed out: we can decide later to group tools
if it would make sense from a management POV.


If you want to create a new Commons component you
> can create a new git repo any time you want to.
>

Thanks.


The only time you
> really need to ask for permission from the PMC is when you want to
> integrate it into the Commons home page.
>

Obviously, this work aims at that. [Otherwise, why bother?]

So, what does the PMC say?


Of course, you should seek
> consensus from your fellow Math developers but if you make a repo and
> populate it I am sure that others will go along if they can see that
> what you have done is a good idea.
>

That's what I did with "Commons RNG".


I keep getting the impression that you believe people are standing in
> your way. They aren’t. If you start a discussion you should expect
> people to weigh in. Just because they disagree doesn’t mean they are
> going to stop you. But if you don’t listen to their opinion don’t
> expect to get much help either.
>

Let's not reverse the roles, please?
Did you read that thread:
  http://markmail.org/message/uiljlf63uucnfyy2
?

Before I ended up being insulted, I helped with "Commons Math"
for an _extended_ period of time despite my long time disagreement
with the management of the codebase.
I spent considerable time doing "consensus" things even when my
opinion had been that they were useless (and proven so, later on).
So, not only did I listen, but I did much more than "listen"...


Gilles



> Ralph
>
>

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

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message