No longer replace `magit-insert-unpulled-from-upstream'
with `magit-insert-unpulled-from-upstream-or-recent' on
`magit-status-section-hook'. The latter is an obsolete
alias for the former now.
... instead of to magit-pull-and-fetch-popup. Most users don't
use the latter popup at all. Also `magit-branch-pull-request'
was updated and is now much better suited for the branching popup.
It's no longer necessary to use `magit-show-commit' to get consistent
behavior in `magit-refs-mode' because as of v2.9 `magit-visit-ref'
defaults to behave just like `magit-show-commit'.
Previously we disabled the margin in log buffer by setting
`magit-log-show-margin' to `nil'. That variable has been
replaced with `magit-log-margin', which also allows showing
just the commit age instead of both the author and the age.
That seems like a good compromise.
The function on `git-commit-finish-query-functions' by default no longer
complains about "too long" summary lines (which is why we set this to
nil before) but it still complains about non-empty second line (which is
a good thing).
The default value of `git-commit-summary-max-length' has been increased
from 50 to 68, which is close enough to the 64 we used here before.
Many users set `indent-tabs-mode' global value to nil. And I agree with
those users, this should have been the default. But the reality is that
it isn't the default.
If you change the default locally, then your default is different from
the "default default", which is a problem because most files that were
authored by people who prefer to use tabs do NOT explicitly set the
file local value to t, because that doing so would (theoretically) be
redundant.
If you do set `indent-tabs-mode' to nil despite this, then you will end
up using spaces for indentation in files whose author decided to use
tabs. What I do instead is to set the file-/buffer-local value for all
files whose author I am. That's a bit inconvenient, but if I didn't do
that then I would have to check every contributions for tabs.
