On 9/1/2011 12:43 AM, David Nadlinger wrote:
If not, I would propose to postpone the voting phase and start the review for another component, e.g. David Simcha's std.regionallocator, now. After Jonas had enough time to finish a version he considers fit for inclusion into Phobos, I would suggest a short (one week) combined review/vote period to decide on inclusion with the standard library.
The std.parallelism review process kind of ended up the same way, by accident/evolution instead of design. I think this is a good modification to our review process: A 2-3 week initial review, then instead of an immediate vote the review manager may postpone the vote until after the next review queue item. This should happen if major changes are suggested and gives the author time to make changes. Then, have a one-week combined final review/vote as you suggest.
One of the things that I found frustrating in the std.parallelism review process was that a lot of (very good) suggestions came in late in the review process when I felt like I had to rush to get them implemented in time for the vote. The more we can mitigate this, the better.
