At 04:09 PM 6/12/00 -0400, Jonathan S. Shapiro wrote:
>I understand the description, but I'm missing something. It seems to me that
>changes often span source files, and that the unit of description you want
>is therefore the "change" rather than the "file".
Well, the file-spanning set of deltas is usually the motivation, and is documented as a single "change" ~= cvs checkin. When you want to see what changes you have in a system, or if the change was correctly applied to all of the source files it hit, then you want to look along the change dimension.
When you want to look at the history of a particular file, to find out when a bug was introduced for example, then you want to look at the changes which hit that file. (If you want to back out a change because of a bug, then you will want to back it out of all the files it hit.)
Or am I completely misunderstanding your question?