If non-nil then Spacelpa repository is the primary source to install
a locked version of packages. If nil then Spacemacs will install the lastest
version of packages from MELPA. (default nil)
The default value will be true when we bust all the major bugs.
New variable configuration-layer-elpa-subdirectory set in .lock file.
Now users creating a branch based on develop or master won't reinstall
all the packages in a new elpa directory.
Use const input for function configuration-layer/elpa-directory
New const variables used as input:
- configuration-layer--elpa-root-directory
- configuration-layer--rollback-root-directory
Fix computation of elpa directory by using expand-file-name.
Fixes#9805
New dotfile variable dotspacemacs-verify-spacelpa-archives
Default value is nil for now, users wanting to help testing this feature can
set it to t and delete the directory ~/.emacs.d/.cache/stable-elpa and restart
Emacs to download the Spacelpa archive and verify it.
Update function configuration-layer/stable-elpa-download-tarball to download
signature files as well.
Update function configuration-layer//stable-elpa-untar-archive to verify the
downloaded archive using epg library.
Example of signature files can be found here:
https://github.com/syl20bnr/spacelpa/releases/tag/v0.300
New function spacemacs-buffer/error that gather all the errors and then display
them in the startup buffer under a new list entry `Errors:`.
Add shortcut `e` to go to errors list entry.
New private function configuration-layer//error to wrap spacemacs-buffer/error
and increment the startup error count that is displayed in a red mode-line when
startup process finishes.
Use the official spacelpa ELPA repository.
address: https://github.com/syl20bnr/spacelpa
ELPA repository configuration is now in a file called .lock at the root of the
git repository. Its goal is to setup the ELPA repositories for a given branch
and it should not be modified! The philosophy of Spacemacs is to never fork the
git repository and this lock file has been put in the repo on purpose.
Only the master branch will have a .lock file that points to spacelpa, the
develop branch won't use the stable ELPA repository and will continue to behave
like it always did.
BUT for testing purpose and until the first major version of Spacemacs that uses
the stable ELPA repo is released I push a .lock file to develop branch that
points to spacelpa.
I understand that some people can be annoyed by this so the variable
configuration-layer--elpa-archives can be overriden by putting a setq in your
dotfile in the user-init function like this:
(setq configuration-layer--elpa-archives
'(("melpa" . "melpa.org/packages/")
("org" . "orgmode.org/elpa/")
("gnu" . "elpa.gnu.org/packages/")))
Set it to emacs-version as it is the saner defaults here. When updating the
version of emacs, packages should be byte-recompiled. Thus this is safer to
isolate each package by emacs-version.
The Frame delete bindings SPC F c and C, are inconsistent with the delete window
and buffer bindings, that use: d and D
Changed the dired-other-frame binding from d to O
Change meaningful values of BACKWARD argument to just nil and non-nil.
Get rid of spacemacs--cycle-themes, it was unnecessary.
Create a defvar for spacemacs--cur-theme.
Add the key binding: t
to open helm-themes, without having to use the arrow up key.
Reorder/rename: cycles backward and cycles forward
to: next and previous
Next is listed first, because it's the most common action, and it matches the
order in other transient states.
Change the t argument in the call to spacemacs/cycle-spacemacs-theme, to the
symbol 'backward, to make it clear what the argument does.
Add documentation for the Themes Transient State bindings.
As described in issue #9464, this package was removed from melpa for
not having a GPL compatible license and is causing an error every time
spacemacs starts.
Continuation of commit 8c3cf7c3ae
Fix docs for "Fish shell and ansi-term" and explicitly enable `truncate-lines`.
I was using the suggested hook as well as `(setq-default truncate-lines t)`,
so my fish shell was acting funny because `truncate-lines` was being disabled
rather than enabled.
