On 12/17/06, Carl Karsten <[EMAIL PROTECTED]> wrote: > Paul McNett wrote: > > sheila miguez wrote: > >> On 12/17/06, Ed Leafe <[EMAIL PROTECTED]> wrote: > >>> On Dec 17, 2006, at 11:39 AM, Paul McNett wrote: > >>> > >>>> Quick things, obvious things, or short things (IOW, easily > >>>> reviewed) can > >>>> just go directly to the trunk. > >>>> > >>>> This is how Ed and I have been working for almost 3 years. I don't see > >>>> any problem with this scaling up to more and more active developers. > >>> Well, at some point we may have to use a more librarian-like > >>> approach. But that would not be for the foreseeable future. And those > >>> are the sorts of problems that I would love to have! ;-) > >> You guys are probably more well behaved than the last group I had to > >> deal with who used a subversion repository; holy freaking cow one guy > >> would do everything on the trunk and break everything for everyone > >> else. > >> > >> and you couldn't complain because he was the president. drove me ape-poo > >> insane. > > > > We like to think that common sense prevails around here, but > > occasionally we'll break the trunk. A recent example is when I thought > > it would be slam-dunk to support wxPython 2.7. I initially made all my > > changes to the trunk but lo and behold it broke wxPython 2.6. So I > > reverted my trunk changes, made a branch, and only merged back to the > > trunk when wxPython 2.6 was fully supported again. > > > > I think subversion will scale well into dozens of active developers, but > > if we ever get beyond that we may need something more distributed. Years > > off, most likely... > > > > I am finding myself pushing the envelope. whatever that means. > > Here are the thoughts that go though my head: > "this seems like a good idea, I wonder if it will work?" > code code, test, fix. > "hmm, it seems to work. I wonder if the rest of the gang will like it?" > moment of reason.. > "what if they don't?" > moment passes... > "I'll just commit it and see what happens." > > This has 2 implications: > 1. maybe I shouldn't be committing so whilly nilly. > 2. someone else may take the other path and not commit something useful.
By all means commit something or suggest an idea to discuss it. With more people chiming in their thoughts, it creates a more diversified spread of ideas. None of us come from the same background and therefore might think of a really good idea the others did not. > > Given a goal is most amount of progress with the least effort, I think the > "revert some of carl's hacks" is currently 'good enough" but I don't think it > will 'scale' as well as you do. > For now it works. Don't worry about it. Ed and Paul have recoded some of my stuff. The point is that at you become more immersed in Dabo, they should have to correct your code less...... > I have never had this level of access to something of this nature, and the > 'responcibility' is a bit ..something. > It's really not as bad as you think. The beauty of subversion is that if you mess something up and delete the entire trunk, Subversion has every single copy of it on record > Personally, I would think each dev should get their own branch, and use some > sort of buddy system to have changes reviewed before merged. or maybe some > unit testing... I du no. that has it's pros and cons too. Nah, doesn't work. Merging from a system like this will most likely be hell..... I've always used trunk for main dev branch and branched any major changes/add-on's until they were stable enough to merge back in. > > Carl K > > _______________________________________________ > Post Messages to: [email protected] > Subscription Maintenance: http://leafe.com/mailman/listinfo/dabo-dev > _______________________________________________ Post Messages to: [email protected] Subscription Maintenance: http://leafe.com/mailman/listinfo/dabo-dev
