[cvsnt] Workflow advice saught

[Drew McLellan]

Ian Epperson wrote:

> I'd recommend re-thinking that stance.  We commit often here.  If we want a
> clear definition of code that is an "end product", we tag it.  Note that the
> longer you prevent a developer from committing, the greater the likelihood
> of having to merge code, as well as greater uncertainty that other
> programmers' changes may be incompatible.  For having multiple programmers
> making lots of fast changes to a project, rapid commits are essential.

I agree that it's good to commit as soon as possible, for the reasons 
you have stated. However, forcing a commit at 1730 is almost arbitrary 
and will cause big problems.

An buggy commit doesn't just break one small corner of the project. 
Essentially everyone's work is core. If your code breaks, everyone stops 
work until you fix it. This is the situation I'm trying to work around.

Thanks for your reply.

Drew