Look at how many times in the last two weeks that I have been able to make time to comment on review board versus how much code I have looked at without it. Without review board, I just can't get to most review requests lately and so it is all or nothing in terms of getting code commentary out of me.
On Tue, Dec 13, 2011 at 3:04 PM, Jake Mannix <[email protected]> wrote: > On Tue, Dec 13, 2011 at 10:41 AM, Grant Ingersoll <[email protected] > >wrote: > > > > > On Dec 13, 2011, at 12:44 PM, Jake Mannix wrote: > > > > > You mean other than a web UI to see the patch, without having to > download > > > it, make sure you have a clean checkout to apply it to, then fire it up > > in > > > your IDE, again making sure you have actually caught all the diffs? > > > > Are you saying it automatically applies the patch and runs the tests? > Now > > that would be useful! If not, what's the time saving other than for > quick, > > on the run feedback for superficial things? Otherwise, don't you kind of > > have to do those steps anyway to know the tests pass? Or perhaps you > have > > a compiler + JUnit built into your brain? Because that's the > functionality > > that takes the most time and once you've done those steps you can just as > > well view the diffs in your IDE. > > > > No the point is that you can keep reviewing the code without having run the > tests, make progress until it looks like you would like to feel it's ready, > and then only at the last minute you or someone else can run the tests. It > also lets people who are *not ever* going to download the patch and run the > tests the ability to easily comment on things which need to be changed in > the patch. > > > > Does "ship it" then apply and commit the patch? Or do I still have to do > > all that stuff above anyway? > > > > No, it doesn't integrate with RCS. Internally at Twitter, we have git > hooks set up such that you do "git review publish" which creates a review > of your current branch against master, and then once you've gotten the > requisite ship-its, "git review submit" which closes the review and merges > your branch into master and pushes it back to origin. > > > > Still skeptical but willing to be convinced, > > > > So ReviewBoard doesn't do anything magical linking to unit tests or svn or > mvn. It would be nice, and someday I'm sure someone will hook those. As > it is, it's most useful in these contexts as higher visibility. You may be > the main person reviewing some code, and you've got it in your IDE and are > making lots of comments, running tests, etc. I, as an interested (but not > *that* interested) party can watch along and follow what's happening in the > code, and then even make helpful comments in-line without having to really > jump in and set up yet another branch directory or git branch and apply > patches and be Involved. > > -jake >
