>>>>> "Nick" == Nick Alcock <[email protected]> writes:

> Currently, `magit-log-edit-commit' erases and buries the commit log buffer, 
> then
> proceeds to do things that rely on the `magit-git-dir'.  This function isn't
> going to work as expected unless the buffer that is current after 
> `bury-buffer'
> is located under the `magit-git-dir' -- but there is no guarantee of that.
> Indeed there is no guarantee that any such buffers exist (they may all have 
> been
> deleted).  Given the semantics of `bury-buffer' it is sheer luck if you end
> up in the right place, and it is quite likely that `magit-git-dir' will simply
> crash.

> The solution is to record the *magit* buffer itself (not its name, the buffer
> itself) in a variable local to the commit log buffer, so that it is guaranteed
> that we always have a suitable buffer to switch back to even if it's been
> killed, then switch back to it at `magit-log-edit-commit' time.

Applied, thanks.

Yann.

-- 
Major flaws in government arise from a fear of making radical internal changes 
even though a need is clearly seen.

  -- Darwi Odrade

Reply via email to