From 5bcd0cee0fc5eba81d254cba91459ba340c71dd3 Mon Sep 17 00:00:00 2001 From: Stefan Kangas Date: Thu, 8 Dec 2022 18:00:19 +0100 Subject: [PATCH] * etc/USE-PACKAGE-NEWS: Delete file. Ref: https://lists.gnu.org/r/emacs-devel/2022-11/msg01535.html --- etc/USE-PACKAGE-NEWS | 277 ------------------------------------------- 1 file changed, 277 deletions(-) delete mode 100644 etc/USE-PACKAGE-NEWS diff --git a/etc/USE-PACKAGE-NEWS b/etc/USE-PACKAGE-NEWS deleted file mode 100644 index c499820755f..00000000000 --- a/etc/USE-PACKAGE-NEWS +++ /dev/null @@ -1,277 +0,0 @@ -# Changes - -## 2.4.4 - -This release prepares for inclusion to GNU ELPA and includes no other changes - -## 2.4.1 - -This is mostly a bug-fix release: - -- Update the documentation for :custom as per #850 - -- Fix broken test due to #850 - -- better tests - -- add test for #845 - -- Support keymap symbol in bind-key. Fix #845 - -- use-package-core.el: use the Emacs set-default function to avoid saving :custom vars twice - -- Fix Travis - -- typo, should be a vector, not a bytecode object - - Solves https://github.com/jwiegley/use-package/issues/842 - -- Add special value back again, in case needed for backwards compat - - I don't know why this special value exists, but perhaps old client code uses it. - - The additional `t' in the macro expansion is accidental but not harmful I guess. - -- Even when there's no :config, run any pre/post config hooks - - i.e., following the existing docs for use-package-inject-hooks, these hooks are - run: - - use-package--foo--pre-config-hook - use-package--foo--post-config-hook - - This should make config customisations more predictable (for example, spacemacs - uses these hooks extensively to allow 'layers' to be customised). - - I got rid of the "special" default value for :config, because it doesn't seem to - be treated any differently than nil. - - Fixes #785 - -- Clarify the documentation for :after - -- add table of contents to README - -- Fix typos - - Typos found with codespell. - -- Fix typos - -- Attempt to explain omit "-hook" better - -- Update tests - -- Switch from `require' to `load' + `featurep' - -- Use `require', not `load', when byte-compiling - -- Make custom-face evaluate elisp. - - Fix #696. - -- Add a line of documentation for (use-pacakage ... :hook). - -- Fix typo in README - -- Fix documentation for defer - -- Add no-query option for pdf-tools-install - -- Fix typo in README - -- Fix all notes in README - -- Mention use-package-ensure in README - - Without requiring `use-package-ensure`, setting `use-package-always-ensure` - did not actually work for me. - -## 2.4 - -### Breaking changes - -- `use-package` no longer requires `diminish` as a dependency, allowing people - to decide whether they want to use diminish or delight. This means that if - you do use diminish, you'll now need to pull it into your configuration - before any use of the `:diminish` kewyord. For example: - - ``` elisp - (use-package diminish :ensure t) - ``` - -- Emacs 24.3 or higher is now a requirement. - -- The `:defer-install` keyword has been removed. It may reappear as an add-on - module for use-package in a future release. See issue #442 for more details. - -- There is no longer a `use-package-debug` option, since `use-package-verbose` - already has the possible value of `debug`. - -- The ordering of several elements of `use-package-keywords` have changed; if - you had previously customized this (or were an extension author adding to - this list), you may need to rework your changes. - -- For extension authors, `:commands` should no longer be propagated down for - autoloading. See more below. - -### Other changes - -- Upgrade license to GPL 3. - -- If `use-package-verbose` is set to the symbol `debug`, any evaluation errors - during package configuration will cause a complete report to be written to a - `*use-package*` buffer, including: the text of the error, the `use-package` - declaration that caused the error, the post-normalized form of this - declaration, and the macro-expanded version (without verbosity-related - code). Note that this still does not help if there are parsing errors, which - cause Emacs to register a Lisp error at startup time. - -- New customization variable `use-package-deferring-keywords`, mainly intended - for use by extension packages, indicates keywords that, if used without - `:demand`, cause deferred loading (as if `:defer t` had been specified). - -- The `:ensure` keyword now accepts a specific pinning sub-keyword. For - example: - - ``` elisp - (use-package foo - :pin "elpa") - ``` - - This ensure the package `foo` is installed from `"elpa"`. - - ``` elisp - (use-package foo - :ensure bar - :ensure (quux :pin "melpa")) - ``` - - This says that `foo` ensures that `bar` is installed, as well as `quux` from - `"melpa"`. It does *not* ensure that `foo` is installed, because explicit - `:ensure` keywords were given. - -- New `:hook` keyword. - -- New `:catch` keyword. If `t` or `nil`, it enables (the default, see - `use-package-defaults`) or disables catching errors at load time in - use-package expansions. It can also be a function taking two arguments: the - keyword being processed at the time the error was encountered, and the error - object (as generated by `condition-case`). - -- New keywords `:custom (foo1 bar1) (foo2 bar2)` etc., and `:custom-face`. - - NOTE: These are only for people who wish to keep customizations with their - accompanying use-package declarations. Functionally, the only benefit over - using `setq` in a `:config` block is that customizations might execute code - when values are assigned. If you currently use `M-x customize-option` and - save to a settings file, you do not want to use this option. - -- New `:magic` and `:magic-fallback` keywords. - -- New `:defer-install` keyword. - -- New customization variable `use-package-enable-imenu-support`. - -- New customization variable `use-package-hook-name-suffix`. Any symbols named - in `:hook`, or in the CAR of cons cells passed to `:hook`, have this text - appended to them as a convenience. If you find yourself using this keyword - to add to hooks of different names, or just don't want such appending done, - you can change the text to an empty string. - -- New customization variable `use-package-compute-statistics`, and an - accompanying command `M-x use-package-report`. See the README for more - details. - -- Allow `:diminish` to take no arguments. - -- Support multiple symbols passed to `:after`, and a mini-DSL using `:all` and - `:any`. - -- `:mode` and `:interpreter` can now accept `(rx ...)` forms. - -- Using `:load-path` without also using `:ensure` now implies `:ensure nil`. - -- `:bind (:map foo-map ...)` now defers binding in the map until the package - has been loaded. - -- Print key bindings for keymaps in `describe-personal-keybindings`. - -- When `use-package-inject-hooks` is non-nil, always fire `:init` and - `:config` hooks. - -- Documentation added for the `:after`, `:defer-install`, `:delight`, - `:requires`, `:when` and `:unless` keywords. - -- `:requires SYM` is subtly different from `:if (featurep SYM)`, in that it - happens before the `:preface`. This means that using `:requires` will cause - definitions in the `:preface` to not be visible to the byte-compiler, - leading to possible warnings about unknown functions, or functions that may - not be available at run-time (which can generally be ignored, since - `:requires` is intended as a check for basic system functionality; `:after` - should be used to check for the presence of other modules). - -- New undocumented (and currently experimental) keyword `:load` may be used to - change the name of the actual package loaded, rather than the package name, - and may even add other names. For example: `(use-package auctex :load - tex-site)`. This keyword is used internally to generate the `require` for a - package, so that deferral is simply a matter of not generating this keyword. - -- The source code is now broken into several files, so that certain optional - features (diminish, delight, ensure) may be maintained separately from the - core functionality. - -- When using the `:after` keyword, now even autoloadeds keybinding are - deferred until after that other package has loaded, in order to allow - convenient `:bind` to maps only present in that other package. Consider the - following: - - ``` elisp - (use-package helm-descbinds - :load-path "site-lisp/helm-descbinds" - :after helm - :bind ("C-h b" . helm-descbinds) - :init - (fset 'describe-bindings 'helm-descbinds)) - ``` - - The binding of `C-h b` here will not occur until helm is loaded; and after - it is loaded, `helm-descbinds` itself is not loaded until the user presses - `C-h b`. - -- For extension authors, if you add a keyword to `use-package-keywords` whose - presence should indicate deferred loading, please also add it to - `use-package-deferring-keywords`. Note that this is a bit of a sledgehammer, - in that the mere presence of these keywords implies deferred loading. For a - more subtle approach, see the new `use-package-autoloads/` support - mentioned in the next bullet. - -- For extension authors, if you wish deferred loading to possibly occur, - create functions named `use-package-autoloads/` for each keyword - that you define, returning an alist of the form `(SYMBOL . TYPE)` of symbols - to be autoloaded. `SYMBOL` should be an interactive function, and `TYPE` the - smybol `command`, but this functionality may be extended in future. These - autoloads are established if deferred loading is to happen. - -- If you specify a lambda form rather than a function symbol in any of the - constructs that *might* introduce autoloads: `:bind`, `:bind*`, - `:interpreter`, `:mode`, `:magic`, `:magic-fallback`, and `:hook`: then - deferred loading will no longer be implied, since there's nothing to - associate an autoload with that could later load the module. In these cases, - it will be as if you'd specified `:demand t`, in order to ensure the lambda - form is able to execute in the context of the loaded package. - -- For extension authors, there is a new customization variable - `use-package-merge-key-alist` that specifies how values passed to multiple - occurrences of the same key should be merged into a single value, during - normalization of the `use-package` declaration into a proper plist. The - default behavior is to simply append the values together (since they are - always normalized to lists). - -### Bug fixes - -- Repeating a bind no longer causes duplicates in personal-keybindings. -- When byte-compiling, correctly output declare-function directives. -- Append to *use-package* when debugging, don't clear it. -- Don't allow :commands, :bind, etc., to be given an empty list. -- Explicit :defer t should override use-package-always-demand. -- 2.39.2