yade-dev team mailing list archive

[Bruno Chareyre <bruno.chareyre@xxxxxxxxxxxxxxx>]

Hi Jérôme,
This is an interesting case, thank you for opening the discussion. In
the past years potential contributors were usually applying a strong
self-censorship, as they were so afraid to break something (so many of
them never commited anything for this reason). As far as I remember it
is the first time that someone asks for more freedom to commit whatever
he likes. So, for the first time, I will try and give a few guidelines
below.

On 03/12/14 23:34, Jerome Duriez wrote:
> Entering now general discussion, do you mean each commit should be
> discussed on yade-dev ?
No. See my messages again, they refer to specific (e.g. "low level")
parts of the code.
> Up to now I considered that thanks to git, the buildboot (when it
> works), etc.., we can enjoy a collaborative work on a (open source)
> code that belongs to everyone, without too much control. Errors, still
> possible, are spotted and corrected thanks to feedback of other devs,
> or users, when errors appear (which, finally, does not happen so often
> it seems).

We are 19 members in yade-dev. Did you ever try building a house or
something with 19 asynchronous workers, overlapping each other but not
talking, without a global plan and with the only feedback "until now it
didn't fail"? I doubt it can work.
It is the same here. If 19 people feel free to push changes to every
files, and without a need to notify what they do, reviewing will become
a full-time job and everything will be broken in a matter of months.
Errors appear not so often precisely because of the self-censorship
mentioned above. Also because a few of us are keeping track of what
happens. It takes a lot of time already, no need to make it more time
consuming.
I do not mean that some parts of the code must not be changed, only
asking for discussion before changing them.

The capillary law is typical. It has no regression tests (then the
buildbot is helpless) and only a few users (delayed and/or little user
feedback - counting on regular users to detect bugs is bad practice
anyway, it refrains them from updating, leading to no user feedback
ultimately).
In my mental map of yade the capillary code is something stable that
nobody is touching. Hence if in 6 months somebody has strange errors
with "yade -j6", I will not suspect the capillary law if I don't know
that somebody (you) modified it. Fixing bugs is time consuming when you
don't know where they are. It is even worst when you think they are not
in a place where they actually are.

>
> If we have to adopt a centralized workflow, I would conform the
> decision whatever my thoughts. But this should come from a global
> consensus, not from a discussion between 2 people, I think.

We don't have to adopt anything new. We reached the consensus in 2011
[1], and it has been confirmed again during the first yade workshop in
Grenoble.
Therefore, I suggest:

Guidelines:
These are situations when we don't expect prior discussion (though it
never hurts):
- commit by a seasoned[2] dev
- change in some code by the author or co-author of the code
- or the notorious maintainer if the original authors left
- or someone following an agreed workplan, in which case we don't need
one email per commit if the task will involve many commits on a long
period (note that git branches are well suited in this situation, though)
- small-scale changes and typos in the documentation or in example
scripts (I know you did that a lot recently, thanks a lot for that)
In other cases it should be discussed.
In every cases, please, put a few words in the commit logs (not like [4]).

Cheers.

Bruno

[1] https://lists.launchpad.net/yade-dev/msg07310.html
[2] where "seasoned" must be considered with respect to what is being
modified. For instance you Jérôme are not yet seasoned in openMP
programming according to [3]. I am not seasoned in debian packaging. I
don't think anyone can claim expertise on 100% of the code actually.
Everyone has to know his own limits.
[3] https://lists.launchpad.net/yade-dev/msg11722.html
[4]
https://github.com/yade/trunk/commit/6c2eb71382d44b6884fb1aa15fdc37af20959c40,
giving the hyperlink is great but it says nothing in itself. We need
something in plain english, then a hyperlink when possible.


>
>
> ------------------------------------------------------------------------
> *From:* Yade-dev
> [yade-dev-bounces+jerome.duriez=ucalgary.ca@xxxxxxxxxxxxxxxxxxx] on
> behalf of Bruno Chareyre [bruno.chareyre@xxxxxxxxxxxxxxx]
> *Sent:* December 3, 2014 1:55 PM
> *To:* yade-dev@xxxxxxxxxxxxxxxxxxx
> *Subject:* Re: [Yade-dev] [Branch ~yade-pkg/yade/git-trunk] Rev 3550:
> Add of O.thisScene() function to know in which scene we are
>
> Jérôme, I think the commit is right but I have the same comment as
> before: please don't touch this kind of code without a note here on
> yade-dev.
> Of course you can change whatever you like in your local copy, so my
> request should not block your production at any point.
> Pushing to trunk is another thing. I can't track changes if people
> start changing low level code silently just like that.
> Thanks
> Bruno
>
>
>
> On 01/12/14 20:49, noreply@xxxxxxxxxxxxx wrote:
>> ------------------------------------------------------------
>> revno: 3550
>> committer: jduriez <jerome.duriez@xxxxxxxxxxx>
>> timestamp: Mon 2014-12-01 11:20:51 -0700
>> message:
>>   Add of O.thisScene() function to know in which scene we are
>> modified:
>>   py/wrapper/yadeWrapper.cpp
>>
>>
>> --
>> lp:yade
>> https://code.launchpad.net/~yade-pkg/yade/git-trunk
>>
>> Your team Yade developers is subscribed to branch lp:yade.
>> To unsubscribe from this branch go to https://code.launchpad.net/~yade-pkg/yade/git-trunk/+edit-subscription
>>
>>
>> _______________________________________________
>> Mailing list: https://launchpad.net/~yade-dev
>> Post to     : yade-dev@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~yade-dev
>> More help   : https://help.launchpad.net/ListHelp
>
>
> -- 
> _______________
> Bruno Chareyre
> Associate Professor
> ENSE³ - Grenoble INP
> Lab. 3SR
> BP 53
> 38041 Grenoble cedex 9
> Tél : +33 4 56 52 86 21
> Fax : +33 4 76 82 70 43
> ________________
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-dev
> Post to     : yade-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~yade-dev
> More help   : https://help.launchpad.net/ListHelp


-- 
_______________
Bruno Chareyre
Associate Professor
ENSE³ - Grenoble INP
Lab. 3SR
BP 53
38041 Grenoble cedex 9
Tél : +33 4 56 52 86 21
Fax : +33 4 76 82 70 43
________________