G. Patch Writer Experience
The contributions to WebCore during the period studied came from 496 individuals among which 283 developers filing
95% of patches (submitting 5 patches or more). Considering our top five organizations, we identified that WebCore patches were submitted by 50 developers from Apple, 219 individuals from Google, 20 BlackBerry developers, 16 developers from Intel, 10 from Igalia and 181 developers come from other organizations.
Noticing good contributor diversity in the WebCore community,we wondered if patches from certain developers
have higher chances of being accepted. In order to assess whether developer experience influence review timeliness and
acceptance, we performed a similar procedure (as described in IV-F) of calculating the number of submitted changes for
each developer and then discretizing patch owners according to their contributions.
We achieved similar results in the differences of response time for A and B groups of submitters (occasional contributors)
is statistically significant compared to more experience developers in C or D groups. From Table IV we conclude
that more experienced patch writers receive faster responses (with median in group D being 48 minutes and mean being
478 minutes) compared to those who file fewer patches (the median for time in group A is 102 minutes and 682 minutes
for the mean).
Investigating the impact of developer experience on positivity of the outcome, we found correlation between two variables
(χ2 (3)=17.93, p-value < 0.01). In particular, statistical difference was found between group A (least active developers) and groups C and D (more active developers) with the median positivity values being 1.0, 0.73 and 0.81 respectively, as
well as group B (less active developers) compared to the group D (most active ones) with the median positivity being
0.63 and 0.81 respectively. This findings suggest that the WebKit community has a positive incentive for newcomers
to contribute to the project as first-patch writers (i.e., group A with the median number of patches submitted = 1) are likelyto get a positive feedback. For the developers of group B (where contributions range between 3–6 patches) it is more challenging to get their patches in, while contributing to the project comes with the improved experience of landing patches and as a result with more positive outcomes.
Our findings show that developer experience plays a major role during code review. This supports findings from our previous work, where we have seen faster response time for core developers compared to the casual contributors on the
project [6]. This appears to show that active developers are being rewarded with both faster response and more positive
review outcome for their active involvement in the project.