[SciPy-Dev] scipy.stats

Travis Oliphant oliphant@enthought....
Tue Jun 1 18:33:56 CDT 2010


On Jun 1, 2010, at 1:57 PM, Matthew Brett wrote:

> Hi,
> 
>>> Well - but that is because you don't maintenance.   Imagine a
>>> maintainer puts in a lot of effort to make the code well-documented
>>> and tested.  Then, you have put in new code that has neither
>>> documentation nor tests.    As a good maintainer, it's really painful
>>> for them that there's new code without documentation or tests.   They
>>> can only feel abused in that situation, because it seems as if you are
>>> expecting them to clean up after you - without asking.
>> 
>> I don't think that is fair.  I have been "maintaining" SciPy and NumPy code for over 10 years.   I have done an immense amount of work in porting SciPy to NumPy and continuing to fix bugs that I am made aware of.  I don't have as much time to commit to SciPy as I would like.
> 
> I wasn't really saying whether it was fair or not, I was only trying
> to explain why it might cause offense.
> 
> When I say that you don't do maintenance, I mean that you are not
> currently the person who has to make sure that the code is readable
> and maintainable.   That is hard and often thankless work.
> 
> I presume that you agree that numpy and scipy code should have
> documentation and tests.    I presume also that when you commit code
> without documentation or tests, that you do not usually intend to come
> back and do these later - say - before the next release.   That means
> that someone else has to do it.  It will take them a lot longer than
> it would take you because they don't know the code as well.
> 

No, that is actually not what I imply but checking something in to the trunk.    I plan to submit tests and docs before the next release when I commit code.    I don't expect anyone else to do that for me.   I always welcome help, but I don't expect it.  

I really think this is more about how people view commits to the trunk than anything else.   I like to use SVN as a version control system.   My commits to trunk are always more incremental.   I like to get things committed in self-contained chunks.   Adding the requirement to put in documentation and tests before committing stretches out that "incremental" work element to longer than I ever have time for in one sitting. 

Clearly, if I were using DVCS to a published branch that could be then merged to the trunk this problem would not have arisen.   I see that I need to move to that style.    People are reading far more into my committing to trunk than I ever meant to imply.    

-Travis



More information about the SciPy-Dev mailing list