Stephen Leake <[EMAIL PROTECTED]> writes: > Well, that's one way to do it. It won't map well to xmtn.
You're trying to map the back-end tool to DVC, and that's wrong. DVC should help you to use the tool. If the tool promotes a flow, DVC should also help you to follow this flow. At the moment, I'm mostly using git. I'm using it because I have chosen it, after evaluating several VCS, I found this one to be mostly better than the others. That's subjective, other people would have chosen bzr, hg, monotone, or whatever with the same approach. Now, if you want to totally unify the different tools, what is the benefit of chosing a tool or another? If I want to use git, please, let me benefit of git's advantages. If other people want to use X, whatever is X, please let them benefit of X's advantages. > If instead we have a good way to show all three diffs in one UI > (which I think we will), then we only need one DVC command, that > runs git-diff three times. That will be totally confusing for the user. 1) "git-diff HEAD" tells you what you'd commit with "git-commit -a" 2) "git-diff --cached" tells you what you'd commit with just "git-commit" 3) "git-diff" shows you the changes you have in your working tree, that are not yet added to the index. I can very well imagine cases where you'd want 2) and 3), but I really don't see why one would want the 3 diffs at the same time. >>> Or extend "diff table of contents" to really mean "status summary", as >>> I thought it did initially. Which would probably lead to not having a >>> separate dvc-status-mode. >> >> That's what I still think. > > Actually, I like Christians's summary, which leads to three modes, > each much smaller than the current diff-mode. I don't think you said whether you'll allow filetree-oriented operations from this summary (mark files, add, remove, commit). If you allow them, you'll end up with the same as the current diff-mode, just splitting the buffer in two. You'll have the same problem whenever showing a diff between two past revisions (if you show the summary, you allow the user to commit, ...). You seem to be investing a lot of time on this, but I still don't understand the _problem_ you're trying to solve. OTOH, I have pointed out several problems in your approach, which are still unanwsered. -- Matthieu _______________________________________________ Dvc-dev mailing list [email protected] https://mail.gna.org/listinfo/dvc-dev
