commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilles <gil...@harfang.homelinux.org>
Subject [Math] Disagreeing about how MATH-1138 has been handled (Was: [math] GitHub vs ASF Repo)
Date Fri, 17 Oct 2014 14:23:43 GMT
Hi.

On Fri, 17 Oct 2014 10:46:53 +0200, Luc Maisonobe wrote:
> Hi Hank,
>
> Le 16/10/2014 20:20, Hank Grabowski a écrit :
>> OK.  I submitted the pull request yesterday.  I'm going to now 
>> remove the
>> diff from JIRA.
>>
>> https://github.com/apache/commons-math/pull/2
>
> Thank you. I have merged this request and pushed the result to our 
> main
> repository. The only changes I introduced were fixing end of lines in
> the new Akima spline files (main and test). Perhaps you should check 
> the
> git setting core.autocrlf on your side.
>
>
> It seems to me this pull request did not make it to our dev list. Did 
> I
> simply miss it or is there a problem in the GitHub setting since we
> updated our repo? Did someone else see the request? If nobody saw it, 
> I
> think we should ask infra to fix the settings.
>

I didn't see the request.

I also did not see the changes before they were committed.[1]

I have no problem with the principle of dropping broken code; but I 
have
one with figuring out when it is okay to do so without notice, ignoring
that care be taken with such changes.

I had suggested to not touch the existing classes and that they should
be first deprecated, and then removed. Since several alternatives for
implementing the functionality were proposed, it would have been 
sensible
to have an agreement on how to fit them within the library (for 
example:
an abstract base class and concrete subclasses for each kind of 
spline).

In CM, we've had, on one hand, small, trivial, changes that generated a
lot of unwarranted heat and stalled for days or weeks. And on the 
other,
here is an example where big changes are pushed without a discussion.

When I dare to make a suggestion about something,[2] it means that
I took some time to think about the proposal; the minimum of respect
for this commitment is to acknowledge the existence of such comments
and provide an explanation as to why it is better to not follow the
suggested path:

   http://markmail.org/message/tjengf3t6j3hqyph

[If alternative views are really so different that a compromise cannot
be reached, it is quite simple to count the people who have expressed
their preference from a list of alternatives (as Phil often posts).
In this instance, only I have expressed my preference; hence I do not
understand why something else has been committed.]

My opinion is that we should have created new classes containing the
working implementation(s) of the interpolation, and deprecated but
kept the old ones at least up to release 4.0, advertizing (in the
release notes and in the Javadoc) that they are not working properly
(although they follow  reference "such and such"). [Someone might
have used that window of opportunity to point to the root cause of
the issue.]

So, was there a problem with that approach?

I'm sorry if this naive questioning looks trivial to some of you,
but I'd honestly like to know if this project is team work, and how
it's supposed to work in practice.

I'm also sorry if this rant has been caused by a simple overlook
of the post I'm referring to above. However even if it's the case,
there is a problem.

I hope I'm not being misunderstood[3]: it is great that Hank
could fix CM's spline interpolators.
In this opinion, I'm only concerned with the overall aspect of
contributing to a project that purports to be more that a bunch
of hooks to math functions, and about the design of which people
who have been contributing for some time might have earned (?)
the right to be listened to.[4]


Regards,
Gilles

[1] And I'm also not yet comfortable with looking at large changes
     due to my surely inefficient handling of "git"...
[2] This is already after the self-censorship filter, on issues
     where I know in advance that challenging the adopted view will
     either be ignored or go nowhere... :-}
[3] As is often the case by people who do not carefully read what
     I write in this forum.
[4] Which, I know, is not the same as being heard, and even less
     being agreed with. ;-)


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


Mime
View raw message