On 03/17/2008 10:01:18 AM, Sam Steingold wrote: > Mark A. Flacy wrote: > | Since there will be no reason to change your current behavior, I > don't > | see how it will help with the transition. > > setting dvc-status-no-selection-action to 'error makes the user > eminently aware of the differences between dvc & pcl-cvs and makes him > change his behavior to specify _strictly_ what files he wants to > operate on.
Excuse me? I don't see the change in your behavior to changeset-oriented version control. I see you treating bzr/hg/git as a different implementation of cvs. > > an alternative is to learn from "hg revert" which refuses to operate > on > the whole repo unless --all is specified in a start contrast with "hg > commit" which does not require --all to operate on the whole > repository. And there's a good reason for that behavior. "hg revert" destroys any local changes you have made; it is normally used when you've changed your mind about adding a certain file to your current change-set. It is *rarely* used to undo all the changes you have made. "hg commit" is immediately reversible (hg rollback) with no loss of data (your changes are still present). It should almost always be used to commit everything that you've changed in a single change-set. > specifically, when there is no selection, dvc should ask for > confirmation before operating on the whole repository. > this, of course, could be implemented by the 'confirm value of > dvc-status-no-selection-action. I'm not a DVC maintainer, so my opinion on the matter really doesn't count. If I were one, I would argue strongly against this feature; I would expect helper tools such as DVC to aid me in using the tools in the fashion they are meant to be used and to steer me away from using them in a sub-optimal way. Their opinions on the matter may well mirror yours. -- Mark A. Flacy _______________________________________________ Dvc-dev mailing list [email protected] https://mail.gna.org/listinfo/dvc-dev
