Hi Rob, On Fri, Mar 14, 2008 at 11:07 AM, Robert Godfrey <[EMAIL PROTECTED]> wrote:
> Hi Rajith, > > > > Since this is a major area, could you please share your design ideas on > > the > > list? > > Like create a JIRA and post a wiki page. > > This can start a disscussion where everybody else can comment/contribute > > in > > terms of ideas or code. > > This will ensure that the community feels part of the process and the > > community share collective responsibility. > > > > Not actually sure this is a *major* area :-) It is simply a new > implementation of the matching algorithms. Well code wise this maybe a small area, but performance wise the matching algorithm does play a critical role. > Are you familiar with DFAs? Very well. > > There isn't really to much more to it than as I described above: it builds > a > DFA machine as you add bindings. Sorry to have picked on this. But we need to lead by example. Even if things are dirt simple, perhaps it is important to explain to the community why u think it is a good idea. Whether u have explored other ideas etc.. This sort of thing will engage the community and people might have ideas on how to improve it or perhpas better ideas. More than anything it helps keep the transparency and improves community participation. > I've not finished it yet, but when I have > something ready for people to review I'll post it up. > > Ok, my goal is to eunsure more transparency of, a) what we are planning to do b) why we are doing it? ex bug fix, feature, improvement etc.. (JIRA is best to capture this) c) how we are planning to do it d) and what sort of time frame (Ex M3...) > > > > > I would hold off on the DTX stuff until M4. We need to significantly > > > refactor the Java Broker (another thing I am doing in the background > and > > > will look to commit after we have M3 out of the way) and also put in > > place > > > flow-to-disk. > > > > > Rather than doing this only in the background, I think you should throw > the > > ideas on the list. > > We as a community need to give enough time for people to understand, > > review > > and comment. > > You could continue building your prototype in the background, but the > qpid > > community should provide more visibility on the design process etc.. > > Also I would appreciate if everybody get a chance to discuss/review the > > idea/design and code before you actually make a commit. > > > > This is something we discussed at the F2F where you were present. All I > am > doing is trying to see how easy it is to move the broker to the > architecture > we described there. I believe Rafi took away the pieces of paper with > that > written on :-) > Lets put this on the wiki. This will help engaging the comunity. We didn't have only a few folks attending the F2F. This will also give the community ideas about the direction we are talking. > I'm not going to commit on trunk - don't worry... Since we settled on the > design in our F2F I think the major thing to do is to have a chat about it > when I have a proposal ready... Rob it is not abouting u commiting on the trunk or not. I wouldn't mind u commiting this on the trunk provided we have adequate discussions. We need to build a community. And we need to capture these discussions/ideas in the wiki/JIRA or where ever if we want other folks involved. People should get more context as to why we are doing something and where we are heading. > > -- Rob > > > Regards, > > > > > > Rajith > > > -- Regards, Rajith Attapattu Red Hat blog: http://rajith.2rlabs.com/
