On Sep 16, 2010, at 12:40 PM, Adam Heath wrote: > On 09/16/2010 01:37 PM, Adrian Crum wrote: >> On 9/16/2010 8:18 AM, Jacques Le Roux wrote: >>> From: "Adrian Crum" <adri...@hlmksw.com> >>>> This description of events isn't entirely true. >>>> >>>> David didn't reject Andrew's design, the community in general felt >>>> excluded from the design process. David simply asked that we discuss >>>> the design before code was committed. >>> >>> Yes exactly, thanks for clarifying Adrian, I knew I had left some points >>> behind >>> >>>> The security redesign was the outcome of that discussion. As far as I >>>> know, David and I agreed on the final design, but interest in it fell >>>> off. I ended up being the only person working on it. Since then, David >>>> has included the security redesign in his new project. >>> >>> I tought there were some stumbling blocks, notably when merging your >>> works. >> >> We only disagreed on the workflow. David wanted to commit all the >> changes at once and I wanted to commit them a little at a time. > > Completely brand new code that doesn't touch anything else *at all* can be > committed as a single large chunk. But if you need to alter a bunch of other > stuff scattered all over, separate commits are better. It makes it easier to > verify correctness, and helps in 4 years when you are trying to figure out > why something is broken.
I agree, it is WAY better to have hundreds of small commits with questionable code state in between them. Again though, Adrian misrepresented what I wanted to do, namely implement the ExecutionContext in a branch and once it is complete and the rest of the framework is cleaned up merge that back into the trunk. I suppose you could say the point of the branch was an attempt to collaborate with others, and on that account it worked out beautifully... I've given up entirely on these things in the OFBiz Framework and instead decided a separate project was the only viable way to see it through. -David