Workflow with 'never commit broken changes'

John Arbash Meinel john at
Mon Jun 26 16:29:51 BST 2006

Michael Ellerman wrote:
> On 6/23/06, John Arbash Meinel <john at> wrote:


> It isn't interesting, it's crack. In the sense that once you've used
> bisect to find a bug you will never go back to a VCS that doesn't
> support it. You'll be hooked.
> I think you're right about bzr, it's just not big/complicated enough
> yet for bisect to be that important, and the test suite helps a lot.
> But for something like the kernel, where a change in generic code
> might break an obscure arch configuration that someone tries to boot 3
> months later, it's invaluable.
> Even then, I think it would be good to use on bzr. If you can automate
> the bisection, it should be quite fast, and it allows you to find the
> exact commit that broke something. So rather than saying "something in
> joe blogs' 400 commit foo branch broke x", you can say "commit
> joe at broke x". Which I think is valuable information.

From all accounts, it does seem like something bzr should get in the future.

>> What I was thinking about, though, is that I commit broken code fairly
>> often. I try not to, but it is fairly common that I'm working on feature
>> A, and to fix it, I need to fix feature B, which I want to commit
>> separately, but in the process I don't worry too much that the entire
>> test suite is passing.
>> Or I've been working on my laptop, an I want to move over to my desktop.
>> Robert mentioned this is abusing bzr as a substitute for rsync/unison.
>> He has a point, but I also know that rsync may through away uncommitted
>> changes on my desktop (or even committed changes that I don't have).
> For that case I'd commit, push, then uncommit on both sides. It's a
> bit of a pain, but better IMHO. I notice there's a new bzrtools
> command called "shove" which caters to this use case even better.
>> We never commit things to the mainline which is broken (which I do
>> support). But I have to wonder what broken commits on side branches
>> might do for something like 'bisect'.
> But why the distinction? If things only work on the mainline, that
> might leave you trawling for a bug introduced in a 10,000 line-diff
> branch merge.
>> I realize I'm trying to justify committing broken code (only in a
>> dev/personal branch). But I think there are circumstances that "NEVER
>> COMMIT BROKEN CODE", would be very restrictive. eg. I'm in the processes
>> of changing X to Y, and I've fixed 80% of the cases, but I need to go to
>> sleep. Do I have to wait until I hit 100% to add a new revision?
> Never say never :) But in general I think it's a very good rule to try
> and follow. At the same time there are definitly advantages to
> committing early and often, so it's a trade off.

Yeah. I do in fact try fairly hard not to commit broken code. But I
don't 100% run the test suite, especially when the test suite takes a
long time to run.
> What I think would be good would be a way to tag a revision as
> broken/working, then you could have the bisect tool be aware of that
> and skip them. That'd free you up from committing broken code, with
> the knowledge that you won't pay for it later.
>> I'm not a kernel hacker, so maybe I'm missing some true enlightenment.
>> Creating a dev process around your tools is required to some extent
>> (human beings are generally more flexible than any tool). But
>> ultimately, you should try to build your tools around the process you
>> want to use.
> I think it's mainly a matter of scale, it's just nasty to have the
> kernel deliberately broken at points on its history. There's enough
> bugs as it is without developers knowlingly committing code which
> doesn't work.
> cheers

Yeah. It depends a lot on how many people are going to depend on you not
having broken things. Project-sized workflows are something that are
important to keep in mind.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : 

More information about the bazaar mailing list