commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilles <gil...@harfang.homelinux.org>
Subject Re: [ALL] The Commons Math issue
Date Thu, 13 Apr 2017 12:03:24 GMT
Hi.

On Thu, 13 Apr 2017 11:53:27 +0200, Stefan Bodewig wrote:
> On 2017-04-12, Gilles wrote:
>
>> [Reminder: a big part of "Commons RNG" was developed inside CM and
>> most PMC people did not even know about it (although I was opening
>> JIRA issues all along.  Hence creating a "git" repository is not
>> futile if it can raise awareness.]
>
> By now you've probably learned that people won't look at JIRA issues
> raised for components they don't work on. At least I don't :-)

A priori, I don't have any problem with an individual taking that
stance. [I do it too, because a day is only 24 hours long.]

But then, one is not entitled to claim a say about the issues which
he let pass...

>
>> On Wed, 12 Apr 2017 15:22:23 +0200, Stefan Bodewig wrote:
>
>>> On 2017-04-12, Gilles wrote:
>
>>>> IMO, there is a contradiction in the PMC being both passive (not
>>>> contributing to the overall health of CM[2]) and active (in
>>>> preventing "do-ocracy" wrt the choice of a roadmap for CM[3]).
>
>>> You can count myself into the camp of people who are willing to let 
>>> a
>>> component go dormant if it doesn't get maintained. But I'm unlikely
>>> to go actively looking for unmaintained components.
>
>> a _lot_ of work has been done (since at least 4 years) on the branch
>> that was bound to become CM 4.0.  I'm inclined to think that it
>> deserves more than being thrown away.
>
> I'm not suggesting to throw away anything. All I said was that I'm
> prepared to move unmaintained components to dormant where anybody can
> pick them up again later. You're saying MATH isn't unmaintained, 
> that's
> fine.

Maybe we mean the same thing.
I've indeed said so one year ago. It is still true, if we consider
that "maintenance" comprises the capacity to absorb JIRA issues
(communication with developers and bug-fixing).

> I'm still not sure where you see do-ocracy being prevented. If 
> anybody
> wanted to RM a MATH release, they can do so.

That's quite true, but _I_ did not want to, for the above reason!

Nevertheless, most the CM pieces have a lot of value, and IMO
would become even more valuable as separate components. [Rationale:
separation of concerns, attracting people concerned by a single
"piece" of code.]

However, to my lasting dismay, some people clung to the bad and
outdated pieces too.
In effect, they did stall work on the 4.0 release of CM.

> And to me it looks as if
> nobody was preventing you - or anybody else - from creating new
> components seeded by code taken from MATH (as long as the number 
> doesn't
> get scary, I hear you, Oliver :-).
>
> It seems creating a git repository as the first step may not be the
> preferred approach, though.

It was not the "first step" (cf. ML archive).

>>> I'm not sure we need a roadmap. IMHO if you can identify a viable
>>> subset of MATH you want to maintain as a separate component, then 
>>> you
>>> should be able to do just that. At the same time this shouldn't
>>> prevent anybody else from working on MATH if they really want to.
>
>> Exactly (although the latter did not happen, and it's something for
>> the PMC to take into account when alternative are proposed).
>
> It is probably a lot easier to accept "let's create a new component 
> that
> focusses on X with code seeded from MATH" than "here is a big plan 
> for
> how we want to deal with breaking up MATH".

There can't be a "big plan" because we lost too much expertise to
carry it out safely.
Modularizing the whole of CM (as advocated by Emmanuel) is a heavy
effort that is unlikely to be rewarding. [Even though "back then"
it was a good idea (I was a proponent).]

> I prefer the "small steps"
> approach taken with RNG and NUMBERS.

That's what I've been advocating all along.

>> As you know, this CM issue has created a lot of grievance.
>
>> I do complain that the PMC did not fulfill its role, by not even
>> trying to safe-guard the "Commons" project's integrity.
>
>> I expect the "Commons" PMC to _support_ the people who work here
>> (cf. "git log").
>
> I read you expect the PMC to do something, but unfortunately I don't
> understand what it is that you want the PMC to do. Maybe we are are
> interpreting the role of the PMC differently.
>
> In what way has the integrity of the Commons project been endangered?
> I've seen people fork the code of MATH - which is fine by our license 
> -
> and move to work in a different environment - which is their choice 
> and
> I'm not willing to judge them.

And I think that the PMC has been wrong in passively accepting the
"surprise" fork.
Because it came from _inside_ the community, the PMC would have
been right to demand that a reasonable attempt be made at exposing
the reasons, and at trying to fix issues while preserving the
community.

I was hurt by the fork, and the way it happened.
And I was hurt that the PMC did not see anything wrong in "community
fellows" keeping me in the dark for five months, to work alone on a
doomed project, while they were sneakily setting up an alternative
environment.

> But the code is still here and anybody is
> free to keep working on it. No danger for the Commons project IMHO,
> maybe a danger for the MATH component going dormant which is 
> something
> that may happen to any other component as well when people stop 
> working
> on them.
>
> I've seen you sticking around to work on MATH and keeping the parts
> alive that you care deeply about and finding new contributors that 
> share
> those goal - which is great.

Or stupid...

I understand that most people here do not understand what I'm talking
about, perhaps because they were not involved personally.
Thus we have to face that "Community over code" is an empty statement:
Comforting when not necessary; but when things started to get awry,
no "community" was paying attention (see December 2015 ML archive).

> The PMC has not been standing in the way of RNG or NUMBERS, maybe
> discussions have been taking longer than you'd have wanted.

Yes that's one of the things that prevent "do-ocracy": someone
willing to do the work is stalled by (non-technical) arguments
from someone not intending to back them with actual work (same
reference): I'm still wondering which part of Commons RNG could
make me deserve the treatment I got when I started working on the
"random" package of Commons Math.

> But that's
> what you get inside a chatty community (I'm deliberatly rate-limiting 
> my
> responses :-). The new contributors have been made committers by the
> PMC.
>
> I'm confident the PMC won't stand in the way of creating new
> self-contained components in the future, some members of the PMC may
> quibble over the details, though, and you'll need even more 
> discussions.

And in the meantime other projects do actual work!
[Oops, I should do that too...]

Gilles

>
> Stefan
>


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


Mime
View raw message