RE: [Codestriker-user] Feature request
Brought to you by:
sits
From: Kelly F. H. <kf...@mq...> - 2004-02-10 21:25:02
|
> -----Original Message----- > From: David Sitsky [mailto:si...@us...] > Sent: Tuesday, February 10, 2004 3:22 PM > To: Kelly F. Hickel; Matthew Hailstone; codestriker- > us...@li... > Subject: Re: [Codestriker-user] Feature request >=20 > > Hi David, > > Just FYI, we don't do development this way, nor does any > > development organization that I've been a part of for the last 20 years. > > We do projects in branches, have the branches reviewed, do unit test, > > THEN merge it to the main line of development. We want our developers > > to be able to do a commit at any time (and urge them to do it at least > > daily) to prevent mishaps due to power outages, drive failure, etc. > > This has paid off more than once! The other issue arises when you're > > working on a change that affects several different platforms (aix, > > solaris and windows for instance), you need to be able to commit, so > > that you can check the code out on the other machines to do unit > > testing. >=20 > Hi Kelly, >=20 > I guess the key message here is that with your process, each developer has > their own separate _private_ branch, so even if they do daily commits to > their private branch, there is no impact to the rest of the development > team. With your process, there is still a way once the work has been > completed, to generate a diff that can be reviewed, and then merged with > the main branch. >=20 > With the process described by Matthew, unreviewed code from more than one > developer is being committed into a branch/main trunk, which I think is a > big problem. >=20 > The basic message is unreviewed code should never be committed to a > "public" branch that other developers are using. David, While I (in theory) agree with your last statement, we do have multiple developers working on projects (not always, but often), and so, they do share a branch. They are fully authorized to beat each other with a wet noodle if one of them breaks the other's code (;>), but it does happen. We do the reviews at the end, and sometimes the reviews cause changes, so we need to review it again, requiring a new codestriker topic, instead of just updating the existing one. -Kelly7 >=20 > -- > Cheers, > David |