std.parallelism: voting postponed

[dsimcha]

On 3/20/2011 9:24 AM, Lars T. Kyllingstad wrote:
> A lot of suggestions have been made for std.parallelism in the last week
> of the review, and David currently doesn't have the time to fix all of
> them.  We will therefore postpone the voting, which was scheduled to
> begin on Friday 25 April, and restart the review process at a later date.
>
> David, how much time do you need?
>
> -Lars

Given that a substantial amount of discussion is still going on (as 
opposed to me just having a fixed list of stuff to implement), maybe an 
extra 2 weeks.  If we want to review something else in the mean time, 
it's ok if we put std.parallelism on hold slightly longer.

Reviewers:  I welcome your suggestions, even if they require design 
changes and require substantial effort to implement.  However, if you're 
going to provide such non-trivial suggestions, though, please do so ASAP 
instead of at the last minute.  This will allow time for the following 
important things that there wasn't enough time for with all the 
non-trivial suggestions I was getting at the last minute:

1.  Discussion/refinement/clarification.  The first time something is 
suggested, it's often vague, based on misunderstanding, etc.  If 
something non-trivial is suggested well in advance, alternatives, 
refinements, clarifications, use cases, etc. can be discussed.  If it's 
brought up at the last minute, I'm basically forced to either implement 
it immediately or dig my heels in against it immediately.  This is 
obviously bad.

2.  Proper thought about how to work any changes into the rest of the 
design.  When something is implemented at the last minute, it tends to 
get bolted on instead of fit nicely into the design.

3.  Implementation.  Code that's written at the last minute tends not to 
be the most well-documented or bug-free.