Re: [O] [PATCH] don't always restore previous window configuration?

[Nicolas Goaziou]

Hello,

Matt Price <address@hidden> writes:

> Then I'm sorry for the noise (and also for this aside!).  Magit, or git, or
> some other tool was doing something quite strange and disturbing, and I am
> now seeing these remotes in my repo:
> dup       /tmp/tmp-orgtest/org-annex-testoRSq4r/ (fetch)
> dup     /tmp/tmp-orgtest/org-annex-testoRSq4r/ (push)
> origin  https://code.orgmode.org/bzg/org-mode.git (fetch)
> origin  https://code.orgmode.org/bzg/org-mode.git (push)
> original        /tmp/tmp-orgtest/org-annex-testrgbPft/ (fetch)
> original        /tmp/tmp-orgtest/org-annex-testrgbPft/ (push)

My bad. It seems "test-org-annex.el" messes with "git init". Since
I don't have git-annex command, these tests are usually skipped, so
I hadn't noticed.

> I'm not sure. Until now only Eric and I had commented on this feature.  I
> will say that I will probably never go back to the old behavior, but maybe
> there are people who prefer it. I believe there are 6 possible options for
> org-src-window-setup (though only 5 are listed in the defcustom definition,
> I think that org-src-switch-to-buffer also recognizes 'silently). It may be
> that some of those options only make sense if the window config is restored
> after closing the src buffer.
>
> ~org-src--saved-window-config~ is only used in this function, so instead of
> adding a new variable we could just get rid of an old one :-).

IMO, the only value for `org-src-window-setup' where it might be
meaningful to preserve windows excursion is `reorganize-frame', since it
can disturb a lot the configuration used before editing the block.
However, the docstring disagrees: "When exiting the edit buffer, return
to one window."

Therefore, I suggest to simply do not restore previous configuration in
Org 9.3, and document it in incompatible changes. It will leave time to
test it and complain if something is not right.

WDYT?

Regards,

-- 
Nicolas Goaziou