After studying set-language-environment, it seems to embody all these
lines. And without without it, non-English glyph rendering is very
slow (see https://emacs-china.org/t/org-mode/16918).
This may have other implications, however, so further testing is needed.
Already set in early-init.el, and deferred compilation is auto-disabled
in noninteractive sessions, so only matters to clients that load
early-init.el.
+ Adds doom-emoji-fallback-font-families
+ Adds doom-symbol-fallback-font-families
+ Gives doom-unicode-font higher precedence (and sets it to nil by
default, since Emacs defaults to symbola already).
+ Store custom face settings in psuedo theme, rather than crowding the
user theme (which may be immortalized to custom-file, which we want to
avoid).
+ Now, after-setting-font-hook is guaranteed to fire after *all* fonts
are set, giving you a more reliable interface to add your own
font(set) modifications.
More testing is needed to ensure the patched fonts our ligatures modules
don't conflict with this change.
Produces more helpful (and harder-to-miss) error messages when a hook
emits an error. Also advises run-hook when doom-debug-mode is active, so
errors in hooks (generally, major mode hooks) don't quietly go
unnoticed.
More in line with Emacs' built-in practice of storing a variable's
standard-value in a symbol property of the same name, with the added
benefit of less global state.
We were making the value of doom-first-input-hook permanent-local
instead of doom-first-input-hook itself. That value is normally nil at
this point, which seems to be harmless. Reloading core.el from a script
(while doom-first-input-hook is not empty) results in an error.
Fix it by adding the missing quote present in the other calls.
This could potentially cause stuttering for long lived sessions where
the user takes few breaks, but could yield great performance benefits in
the shorter term. We'll have to experiment.
I was conservative before. I didn't want to change this globally in case
it had a negative performance impact on shy servers that don't talk
much, but this turned out to be too paranoid. Untalkative servers are
barely, if at all, affected, and chatty ones perform better across the
board.
This fixes a class of issues where doom-first-file-hook would fire even
for files not opened interactively (e.g. via org-agenda reading agenda
files or helpful scraping definitions from source files).
In instances like these, hooks like find-file-hook or pre-command-hook
are set to nil to speed up this process and reduce noise, but many
packages add to these hooks when they are loaded; these are lost when
their lexical values fall out of scope, leading to odd errors (like the
one in #4759).
Fixes#4759 (again)
First brought up in hlissner/doom-emacs@9f08db8
Part 2 of 811099282.
The latest build of native-comp appears to eagerly load comp very early,
so defadvice! won't be defined when this block runs, causing this error.
This is commonly used in Emacs packages to resolve to a location in
`~/.emacs.d`, as storage or a cache. Rather than addressing each
file/directory variable as they are encountered, better to address this
at the source.
Also: this makes some file/directory settings redundant, so they were
removed.
Closes#4347
Renamed `comp-deferred-compilation-black-list` to
`comp-deferred-compilation-deny-list`.
Removed the `late` load flag which is no longer required.
Added a check against the deny list when compiling all Elisp on the
load-path, so we don't inadvertently compile something we shouldn't.
Added compatibility shims to ease transition from older builds of
native-comp.
