[TYPO3-core] Review system woes

[Robert Lemke]

Hi all,

On 07.02.2013, at 10:00, Dmitry Dulepov <dmitry.dulepov at gmail.com> wrote:

> Doesn't this mean that the current system is not meant to pass good practical patches through but instead to pass something, which is artificial, theoretical and academic?


with the risk of oversimplifying things, I think that there are two reasons why reviews work like this:

 1) change sets are difficult to review
 2) reviewers use their limited time to reviewing patches they feel more connected to

The first point can (and needs) to be solved by having first-class patches with a descriptive commit message, related tests and basically all information you need to reproduce the problem or test the new feature. Personally, I also think that Github can lower the barrier a lot and makes certain processes more visible.

The second point is more difficult to tackle. I think it's just natural that if you take an hour or so from your evening spare time for reviewing, you'll pick those changes which you like and don't cause too much trouble.

One (additional) way to solve this second point is this: if you customer needs a patch ending up in the core, also add a budget for review into the calculation. It's only fair to pay someone for doing a proper review (and that shouldn't be a guarantee for letting the patch pass of course).

Another idea is a boyscout's approach: review three patches (or so) before you submit your own patch for a customer ...

Cheers,
Robert

-- 
Robert Lemke
Lead Developer TYPO3 Neos and TYPO3 Flow
Co-Founder TYPO3 Association

Blog: robertlemke.com/blog
Get involved: typo3.org – flow.typo3.org – neos.typo3.org