On 2/27/14, 10:10 AM, Craig Dillabaugh wrote:
On Thursday, 27 February 2014 at 02:34:53 UTC, Andrei Alexandrescu wrote:
Unfortunately we won't participate in GSoC this year. The decision was
not surprising - our application has been rejected.

Sadly there are lots of things we could have done better. Our
application has been a low-priority side job for Walter and myself and
as such its quality has suffered greatly.

GSoC applications are a great example of things where one or more
community members can have a large impact on D's well being by
offloading a parallelizable work from the two of us.

Please consider taking a leadership role for GSoC 2015.


How much time did you spend on the application this year?  How much time
do you think would be needed to put together a good quality proposal?

Walter and I pleaded that the other completes the application, with me saying I don't have the time and him saying he's not suited for the job. In the end I "won" and he spent a couple of hours drafting a proposal, which was indeed bad. I spent maybe an hour a late evening trying to improve the proposal and that was about it. Made no page on dlang.org and did nothing on the wiki ideas page (which I think was weak as well).

But sheer time spent is not essential here as the availability of mental cycles. When I do something right I think of it in small quanta all the time - showering, walking, running, whatever. So by the time I sit down to work on it I have ideas and plans already formed. The GSoC was the exact opposite - unprepared "todo" work vying for attention at the periphery of an already overflowing plate. There's no way I could have done a good job at it.

For better or worse Walter and I are the bottlenecks on a lot of D-related stuff. (Just look at http://goo.gl/jGYzir which is developing a nice tenure as a tab in my web browser.) Kenji wrote me an email months ago asking for my take on DIP49, and has done a lot of legwork before I came back to him saying we need a radical simplification. No wonder he wouldn't answer my emails. Whenever anything comes, I need to act "managerial" - absorb context quickly, make a decision, delegate details, move on.

There's just too much important AND urgent stuff going on right now in D, which gives a whole other perspective on the people who advise us on how to do things better, to dissolve into the shrubs when a very concrete opportunity to do something. From that angle, every single little thing that's "parallelizable" and off our plate (such as build system, auto tester, release management, GSoC, and such) is a double improvement for the language as a whole: once because that part gets done better, and twice because it frees us to better focus on other things. Concretely: there wasn't much time to work on allocators lately...



Reply via email to