* Use the terms "Oneline" and "Verbose" to describe the two log types.
Previously "Short" and "Long" was used, and the latter is rather
ambitious and depending on how you look at it simply wrong.
* Make sure the action buttons are arranged in a helpful manner.
* Rename the log commands to better reflect their purpose.
* Move the heavy lifting to the "core" functions instead of
offloading it to the "variations".
And most importantly focus on showing logs for refs and revs, not
ranges. Showing logs for ranges really isn't all that useful, there are
cases when it is, but showing the history of a branch different from the
current branch is VERY useful. That was previously already possible by
abusing the "ranged log" to show a branch (by omitting the end of the
range), but that was VERY confusing to users. Now one has to abuse the
"branch log" to show a "ranged log" and in that case the completion is
less helpful than it used to be. That will change though.
`magit-mode-refresh-buffer' which is used to refresh any magit
buffer (which includes creating it) tries to keep point at the
same section. However it only knows about one buffer at a time.
So if point is e.g. at a commit inside one of the "mini logs" in the
status buffer and the log buffer is brought up, then point is usually at
the beginning of the buffer but may also end up at some random position.
This commit teaches log how to goto the commit that is at point in the
buffer that was current before the log buffer was created. This is done
mostly for logs that are used to select a commit. Support for such
special logs will be added in the next commit. But instead of limiting
it to just these special logs we do this here for all logs. This should
be further generalized so that it can be used in other situations too.
The commands `magit-commit-fixup' and `magit-commit-squash' previously
used some custom code to select a commit from a log buffer. Here we
generalize that to make that feature more flexible and also so that
we can use it elsewhere.