Stephen Leake <[EMAIL PROTECTED]> writes: > In addition, any existing status buffers are cleared, since they are > created with type `diff'. But instead of simply clearing the status > buffer, it would be more useful to only delete the files that were > committed. That would leave any unknown files, for example; very > useful when adding a large existing project to a new DVC backend. It > would also better support partial commits. The point is to avoid > running the backend whenever possible, since it can be slow. This is > also what pcvs does, for the same reason.
Yes, there are many other things that DVC could do to keep status and diff buffers as up-to-date as possible without re-running the back-end. For example, when you save a versionned file, it could show-up as modified in status buffer, ... BTW, shrinking instead of clearing the diff buffer also makes sense for partial commits. I suppose you didn't finish writting your mail before sending it, given the subject line and the end, but we can imagine a kind of buffer-to-buffer communication: each buffer could declare a function in a local variable, like (dvc-files-changed (file-list)), another (dvc-files-commited (file-list)). When commiting a list of files, you'd (funcall dvc-files-commited files), and you'd let the function do the right thing in the buffer in which it's declared. -- Matthieu _______________________________________________ Dvc-dev mailing list [email protected] https://mail.gna.org/listinfo/dvc-dev
