commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Luc Maisonobe <>
Subject Re: [GitHub] commons-math pull request: Updated the example documentation for t...
Date Fri, 03 Oct 2014 11:42:17 GMT
Hi all,

Le 03/10/2014 12:17, Gilles a écrit :
> Hi.
> On Thu, 02 Oct 2014 15:35:53 -0500, Ole Ersoy wrote:
>> On 10/02/2014 02:18 PM, Phil Steitz wrote:
>>> On 10/2/14 11:33 AM, Ole Ersoy wrote:
>>>> Hi,
>>>> On 10/02/2014 11:34 AM, Gilles wrote:
>>>>> Hello.
>>>>> On Thu, 02 Oct 2014 10:51:53 -0500, Ole Ersoy wrote:
>>>>>> Hello Luc, Gilles, and Benedikt,
>>>>>> I'm here :).  From my limited github experience, I do think most
>>>>>> contributors do their communication directly through github.
>>>>> "Commons" contributors?
>>>> I should have clarified.  Just github users and contributors in
>>>> general.  A popular pattern I see for github repository projects
>>>> is that developers keep their pull communication with the pull,
>>>> design (etc.) discussions in issues, and support on
>>>> stackoverflow.  So I'm thinking that future / potential
>>>> commons-math contributors might expect this type of pattern.
>>>>>> I've
>>>>>> seen a lot of projects use issues for discussion.  I personally
>>>>>> like
>>>>>> this, because it makes it easy to get up to speed on design
>>>>>> decisions
>>>>>> and project history.
>>>>> Perhaps it's a better way.  But it's not (yet) the official way,
>>>>> which
>>>>> is this ML.
>>>> Sure.  I just wanted to mention it since github users that are
>>>> watching the project will get more information without having to
>>>> subscribe to the mailing list.
>>>> Also I think what Benedikt was saying was that I might be someone
>>>> who does not actually know about the mailing list, so if
>>>> communication goes out to me on the mailing list WRT my pull
>>>> request, I might miss it.
>>> The problem is that if any substantial design discussion happens
>>> *outside* the ML, those who *are* subscribed to the list will miss
>>> *that*.
>> Sure - but in all fairness, it's as simple as going to the github
>> repository and clicking watch.  The other benefit is that now the
>> notifications will be for commons math only.
> ?
> We _have_ to be subscribed on this ML to contribute to this project.
> Thus we get all the (not necessarily interesting to everyone) posted
> here, and _in addition_ we would get message from yet another source.
> That's no improvement. Unless things have to get to get worse before
> they get better. ;-)
>>> A core principle of ASF projects is that everything
>>> happens in the open, is archived and easily accessible to anyone
>>> interested in getting involved in a project.
>> This will become even more true if github issues are used.  From a
>> reading perspective the github format is cleaner looking than
>> markmail, etc.
> IMHO, the readability of the archive is not the most important point.
> It is there for reference purpose. Questions are: Where is the "official"
> record of the project and where do discussions and decisions take place?
> The current answer is "here".
>>> The "if it did not
>>> happen on the list, it did not happen" principle is really just to
>>> ensure that.  If we start having design discussions outside the
>>> list, to figure out what is going on / has gone on, people will have
>>> to go looking around the internet, rather than just looking at the
>>> list archives.
>> Everything is tied to the repository.  All the information is in one
>> place.  That's usually a plus.
> Only if we drop everything that would be a duplicate of the functionality
> available on that other forum.
>>> This is one reason we have traditionally liked to
>>> have design discussions on the list, rather than in JIRA tickets and
>>> why JIRA comments are in any case forwarded to the list.
>> Sure - honestly I'm just trying to be helpful and suggest a few
>> things that will make things even simpler while improving the utility,
>> uniform accessibility, visibility, and marketability of commons math
>> activity.
> Thanks for trying to help. However, you might have more luck to create
> things that do not exist yet rather than try and change what exists.
> [From my own experience. :-} ]
>> Since git is a more social platform, it may make sense to
>> take advantage of more of its features.
> In practice, you may be right; but in principle, I don't agree.
> "git" provides the means to decentralize while "github" is by essence
> a centralizer.

There are two separate things: discussing features and providing patches.

Discussion must be on the list. This is Apache way of doing things and
it is where people will look at. It is one of the few things that are
mandatory at Apache. Someone said once: "if it didn't happen on the
mailing list, it didn't happen" (most probably one of Apache early

Git and its decentralized aspects improve on how contributors can
provide patches. They can experiment ont their own development setup,
and once they are happy with it they can trigger a pull request. I
sincerely think this is really a good thing and it allows these people
tog et involved more quickly. It also seems more fair to me since Git
allow to have a separate author and committer, so new contributors get
the credit they deserve more efficiently than with subversion, where
only the committer name is automatically preserved.

So you can for sure use GitHub for the pull request, it's great!
However, the mailing list remains the main discussion place.

best regards,

> Best regards,
> Gilles
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message