[Numpy-discussion] NA masks in the next numpy release?

[Scott Ransom]

On 10/23/2011 04:07 PM, Robert Kern wrote:
> On Sun, Oct 23, 2011 at 20:58, Matthew Brett<matthew.brett at gmail.com>  wrote:
>> Hi,
>>
>> On Sun, Oct 23, 2011 at 12:53 PM, Charles R Harris
>> <charlesr.harris at gmail.com>  wrote:
>>>
>>>
>>> On Sun, Oct 23, 2011 at 12:54 PM, Matthew Brett<matthew.brett at gmail.com>
>>> wrote:
>>>>
>>>> Hi,
>>>>
>>>> On Sun, Oct 23, 2011 at 11:21 AM, Nathaniel Smith<njs at pobox.com>  wrote:
>>>>> Hi all,
>>>>>
>>>>> I was surprised today to notice that Mark's NA mask support appears to
>>>>> have been merged into numpy master and is described in the draft
>>>>> release notes[1]. My surprise is because merging it to mainline
>>>>> without any discussion on the list seems to contradict what what
>>>>> Travis wrote in July, that it was being developed as an experiment and
>>>>> explicitly *not* intended to be merged without further discussion:
>>>>>
>>>>> "Basically, because there is not consensus and in fact a strong and
>>>>> reasonable opposition to specific points, Mark's NEP as proposed
>>>>> cannot be accepted in its entirety right now. However,  I believe an
>>>>> implementation of his NEP is useful and will be instructive in
>>>>> resolving the issues and so I have instructed him to spend Enthought
>>>>> time on the implementation. Any changes that need to be made to the
>>>>> API before it is accepted into a released form of NumPy can still be
>>>>> made even after most of the implementation is completed as far as I
>>>>> understand it."[2]
>>>>>
>>>>> Can anyone explain what the plan is here? Is the idea to continue the
>>>>> discussion and rework the API while it is in master, delaying the next
>>>>> release for as long as it takes to achieve consensus? Or is there some
>>>>> mysterious git thing going on where "master" is actually an
>>>>> experimental branch and the real mainline development is happening
>>>>> somewhere else? Or something else I'm not thinking of? Please help me
>>>>> understand.
>>>>
>>>> I don't know about you, but watching the development from a distance
>>>> it became increasingly clear to me that this would happen.  I"m sure
>>>> you've had the experience as I have, of mixing several desirable
>>>> changes into the same set of commits, and it's hard work to avoid
>>>> this.  I imagine this is what happened with Mark's MA changes.
>>>>
>>>> The result is actually an extension of the problems of the original
>>>> discussion, which is a feeling that we the community do not have a say
>>>> in the development.
>>>>
>>>> I think this email might be a plea to the numpy steering group, and to
>>>> Travis in particular, to see if we can use a discussion of this series
>>>> of events to decide on a good way to proceed in future.
>>>>
>>>
>>> Oh come, people had plenty to say, you and Nathaniel in particular.  Mark
>>> pointed to the pull request, anyone who was interested could comment on it,
>>> Benjamin Root did so, for instance. The fact things didn't go the way you
>>> wanted doesn't indicate insufficient discussion. And you are certainly
>>> welcome to put together an alternative and put up a pull request.
>>
>> I was also guessing that something like this would be the reply to
>> Nathaniel's post.
>
> But it wasn't. It was a reply to your message.
>
>> I think this reply is rude because it implies some sort of sour-grapes
>> from Nathaniel, when he is politely referring back to an explicit
>> reassurance from Travis.
>
> What Travis assured did happen, just on the pull request (on which
> everyone's input was requested and where most "should this be merged?"
> discussions are *meant* to happen) rather than on the mailing list.

Except that for a project with a large user community (like numpy), you 
will _not_ get the feedback you are looking for on github pull-request 
pages.  That's because most users do not look at detailed developer 
related things like pull requests.  But they do read the mailing list.

I don't use these features so I don't have a dog in this fight.  But 
potentially controversial changes really should be discussed on the 
mailing list rather than on pull requests (and yes, I know that there 
was a lot of discussion about this stuff some months ago).

Scott



-- 
Scott M. Ransom            Address:  NRAO
Phone:  (434) 296-0320               520 Edgemont Rd.
email:  sransom at nrao.edu             Charlottesville, VA 22903 USA
GPG Fingerprint: 06A9 9553 78BE 16DB 407B  FFCA 9BFA B6FF FFD3 2989