On 06-02-2012 14:28, Jonathan Campos wrote:
Why do you need consensus just to get started? I think this thinking on the
ml is continually crippling are our output.
I started this discussion to see if it was possible to provide
"sandboxes" to each separate initiative, so groups of people can
organize around each initiative. Since there is not a common roadmap and
anyone can have their own idea, many people will be pushing Flex in
their different directions (which is a good thing).
The current model is nice if there is no disruption from what we have
and everyone is creating small features here and there. But I currently
see people wanting to invest time in completely parallel ideas: (1)
Making Flex reach 4.7 (2) Refactoring the framework to make it more
lightweight and modular - Flex 5? (3) Researching and implementing
solutions for exporting the existing SDK to JS/HTML5 (4) Building a new
SDK from scratch specifically designed to target HTML/JS (5) add your
idea here.
Until now, we've seen many people proposing either one of the above
topics, with enough energy to start working on it. Instead of answers
like "well, I might not agree completely, but here you have: share your
vision in this wiki, put the code here, and start working on it (and
remember that there's a risk we might not accept it)", we're just seeing
answers like "No, I don't agree", "Yes, I agree". Let's give these
people resources to start working on their ideas, and to help others who
believe on them work together around the same objective.
Imagine I just subscribed to the ML. I'll start by reading a couple of
emails in the history, but I won't read 2 or 3 thousand emails for sure.
I have an idea for something I would like to work on, so I send an email
suggesting it. I don't have a clue if someone else is already working on
it. In the model I'm suggesting, either:
(1) someone already had a similar idea, so people would simply answer by
indicating the wiki page URL, that would have some info (vision,
roadmap, specifications) on that idea, the respective branch, and a
listing of the persons currently involved
(2) it's a new idea and there's a positive feeling around it so I get
the resources to start my own sub-project. Other people might join later.
If we don't provide tools to help people with disruptive ideas
organizing in groups, they'll simply get those tools outside of Apache
and probably discuss and work outside of Apache, and then bring them in
when they feel it's ready. Not that this is a bad thing, but I would
prefer it if was easier to keep self-organizing groups of persons with a
similar idea as close as possible to Apache.
For example, I have a disruptive idea for Flex that I would like to work
on in a mid-term future. It's a completely different one from what has
been discussed before, and while many people will relate to my vision
and be willing to join, many other might be completely against it. If I
decide to start working on it and others want to join, should I get a
wiki+branch+Jira project+ML outside of Apache (which means, I'm working
outside of Apache...), or should I use the existing resources (that are
all dedicated to one single huge project, so I'll be generating unwanted
noise for all those who don't share my vision) ?
What if we consider a solution that involves providing easily resources
to each sub-project (a codename, access to a part of the wiki, a folder
in the SVN, a JIRA project, and a ML [Tag] (for filtering)) ? This way,
everytime someone joins the ML and says "I want to see Flex exporting to
HTML5", we'll simply say "ok, here's the link to wiki page for that
sub-project, here's where you commit code, here's where you track the
issues, and if you send emails related to that subject add the
[HTML5EXPORT] tag to the subject"...
João Saleiro
