Includes a minor optimization: reading the current level from an
org-element object, rather than relying on (org-current-level), which
resorts to walking the buffer with regex.
Now that 29.1 is stable, support for it is official. This updates or
docs and doctor checks to take this into account. I've also included a
link to a Discourse post where I track support for Emacs HEAD.
Emacs can unpredictably prompt the user to "Select coding
system (default utf-8):" in some cases. The exact cause is a little
different for every user, but it can be suppressed by explicitly setting
a default language environment. This is not desirable in interactive
sessions, however.
Ref: #3042Close: #7330
Co-authored-by: bennyip <bennyip@users.noreply.github.com>
Some psuedo module categories (like :core and :user) don't have a module
component. Rather than display them as ':core nil' or ':user nil' in
module listings (like doom/help-modules), omit the nil entirely.
Some shells (like ksh on SDF) may complain about $((...)) arithmetic
expansion syntax. Rather than wrestle with old shells, I'll offload this
trivial operation to elisp instead.
Close: #6970
From now on, our documentation will assume your Emacs config lives in
~/.config/emacs, by default, rather than ~/.emacs.d. Support for the
latter is not going away, it will simply be mentioned less in the
literature, as all supported versions of Emacs going forward (and future
versions of Doom) will support (and prefer) XDG conventions.
The user manual will be updated separately.
Close: #6965
Co-authored-by: gagbo <gagbo@users.noreply.github.com>
Slipped by manual testing due to consult not being loaded at the time
and doom--help-search going to default grep-find usage instead of
hitting the failiure.
Amend: 54c4340740
`doom doctor` throws a type error for enabled modules without a
doctor.el and/or packages.el file. This occurs because, in
in775ee2f04aad, `doom-module-expand-path` was changed to return nil if
the target file did not exist, and the doctor wasn't updated to handle
this. It passes `nil` to `load`, causing this type error.
Amend: 775ee2f04a
file-name-directory consults this variable for alternative strategies,
which is unnecessary work. (file!) won't (and should never be used in
any context where it could) return anything but a simple file path.
This command would skip over consecutive package! statements in the
generated commit message. This commit fixes that.
That said, this command is still a temporary measure until formal bump
CI/CD is done, but should make dealing with doom/bumpify-diffs and
doom/commit-bumps a little less painful.
When calling backward-kill-word (which is in turn is calling kill-new),
not only kill-ring variable should be overridden, but also
kill-ring-yank-pointer.
Otherwise when deleting a word backwards, kill-ring pointer is moved and
next yank will insert (not so expected) rubbish.
emacs-compat/compat@01fdf316a4 -> emacs-compat/compat@38280a7b54
- Switching to emacs-compat/compat so we can more closely follow the
source. The emacs-straight mirror can sometimes lag too far behind to
promptly address major issues.
- This, combined with 67b3d5a, should address odd byte-code issues to do
with compat (usually to do with magit).
Ref: 67b3d5a20486
This package is cropping up in packages everywhere. Managing it has been
a source of issues, so I'm making it a core package until v3, where
we'll be able to pin packages without explicitly installing them.
Library packages (like compat and transient) offer macros. Therefore,
any package that uses them (dependents) needs to be recompiled when
these are updated, but straight currently doesn't do this. As a
temporary workaround, this commit forces dependents to be rebuilt after
updates.
This is a bit too brute-force, but will do until v3, which will manage
dependency graphs and their complexities more efficiently.
