[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [ide-dev] Development Process

Hmmm…my kneejerk response was that Reviews work best only if people are taking on both roles. That is, the only way to really know a code base well enough to make a sound judgement about code is to also be actively developing to the same code base. It's been my experience that it doesn't work very well when reviewers are not grappling on a day to day basis with the same issues that a contributor is. That might be different for a very mature piece of code, but note that the number of people with those kinds of chops and experience on platform and JDT is very small and it would be a shame to have them doing nothing but reviews. Conversely, I think it is also very important to have every coder involved in reviewing code.

There is a bit of a quid pro quo that has to develop to make the whole exchange work. If we want a community of "social coders" to emerge, we will need to stop thinking about "Reviewing" and "Coding" as separate activities.

(And we need the tools to support that but that's a totally different topic. :))

On 2013-09-25, at 2:07 AM, Wim Jongman <wim.jongman@xxxxxxxxx> wrote:

> Thanks for stating this. Here is a mail from PW a few days ago to e4-dev [1]. Like you said, a committer cannot be  programming and reviewing. Another pattern: the older the patch, the harder to get it in. 
> 
> I agree, it would be great if we had one guy with the skills of the current platform committers dedicated to reviews.
> 
> I clearly remember a quote from Boris Bokowski a while ago (I cannot find it real quick). It was something like "What to do: Do the programming myself or help 10 others with their programming". 
> 
> [1] http://dev.eclipse.org/mhonarc/lists/e4-dev/msg07853.html 
> 
> 
> 
> On Wed, Sep 25, 2013 at 9:07 AM, Mickael Istria <mistria@xxxxxxxxxx> wrote:
> Every committer with the ability to do review are also committed in writing code. To take the example of JDT, all active committers are all working on Java 8 as Stephan explained (which is totally fine), so there is no time left available for reviews. There is a dozen of incoming reviews for JDT.
> So the fact that there is no-one currently available to review stuff slows down the contribution process makes that some of this reviews will get ignored for some long time, and maybe not even make it for Luna. Why? Because reviewers are currently working on writing code.
> 
> The current bottleneck for contributions is not the amount of code produced by contributors, but the ability to review it by committers.
> 
> That's why I still think that getting the Foundation or contributing companier to hire developers won't make more contributions merged. Instead, ensuring there are some people who are committed in reviewing and which make reviews their very #1 task. Such people wouldn't be required to write features or to follow a roadmap. Just they would review contributions whenever one gets in. I don't think a company would be ready to hire someone just to review stuff, but they might be OK to found someone which would significantly speed up the contribution process.
> So the Eclipse Foundation could hire the "Linus Torvalds" of Eclipse, who would commit in reviewing contributions on the Core parts of Eclipse (Platform, UI, JDT). But that brings various questions that doesn't make it so easy: Who is the Linus Torvalds of Eclipse? Can the Foundation afford it?
> _______________________________________________
> ide-dev mailing list
> ide-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/ide-dev
> 
> _______________________________________________
> ide-dev mailing list
> ide-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/ide-dev