configuration VS selectrum

FWIW I have this little Perl script that fetches BibTeX from doi.org for DOI numbers: https://github.com/cadadr/configuration/blob/76466b1342aaadfddf3453ab70ada4a15e82afbb/bin/doi2bib.pl I searched a lot for a way to make something similar for ISBN -> BibTeX to no avail...

If you're writing LaTeX without using any supporting software, it will be harder. But there are some things that you can use to make life way easier. One of these is the concept of "snippets". In Emacs I use something called yasnippet, and it works like this: I define a snippet like this or even something more elaborate like this. They have "trigger words". E.g. for the second one, I type "report" in a file and hit TAB. It inserts all that "snippet" to the file, and I can edit parts marked as $1, $2 and similar, jumping between them using TAB. I use this with Org mode, which is like Markdown, so it's not much different from typing into Word, essentially. There's some initial work figuring out how to do something like a syntax tree or say an equation, but once you figure a pattern out you can make it into a template using snippets and it's easier than Word once you have that.

There are some examples in my init.el, you can find them if you search for :after.

The way I do python is, I've a proxy shell script which I set as the python interpreter in Emacs.

One particular thing I can suggest for equations is dynamic latex equation previews, which toggles TeX source when the cursor is on an equation but when it goes out of it it toggles a rendered preview. See this and this, adapted from this).

Wrt LSP specifically, I only use it with python and the whole config is this couple lines plus this hook. IDK how lsp-mode is configured but this works fairly well for me, and actual Python config is a bit more cumbersome (because Microsoft comes up with a new Python LSP package every other day and you can't know which one to deal with...). It hooks into Emacs' complete-symbol (i.e. C-M-i), so if company or whatever (sorry, I don't really know those packages well, there was auto-something too but IDK which one is better or recommended these days) does use that mechanism as a backend, it should work seamlessly. I've made these little bindings to quickly pick a completion from the *Completions* window (gk-interactively is just a macro that expands to (lambda () ). Again, eglot hooks into Emacs' completion mechanism, so I'd risk a guess that helm or ivy would just work with that. Personally I don't like these completion frameworks because again they look to me like they are somewhat useful but not enough to warrant their complexity. I like good old completing-read, with some modern configuration (and BTW the UI Semantics section that bit is in in my init.el contains a lot of what you could call "saner defaults"). Notably they've added some very neat structural and fuzzy matching abilities starting with 25.1 IIRC and I don't even feel the need to turn ido on when those features are enabled.

I have a little project system that I use a lot and is a life saver: https://github.com/cadadr/configuration/tree/3e11ef25344188cc55b16f314c3c5358ace8a266/emacs.d/init.el#L1249 It's simple but it's very helpful. I like to have a project command view with dired at root on the left and magit or vc.el on the right, and I can go back to that view with a single command, gk-home, bound to the Home key. Popping a shell at bottom like yakuake with a single keybinding to gk-pop-shell is very useful too. I use frame parameters to tie projects up with frames so these two functions know what to do in each project frame. Titles are set up such that it's easy to find a particular frame with something like Rofi.

Depends on what you want to do. Linux is a more welcoming environment for programmers and an OS-level package manager is very helpful. I have one big repo for all my configs, and if my computer failed today I can get up and running on anything else in a couple hours, typing a few commands only. With Windows it's always a manual process and takes longer. And sometimes Windows makes some programmers' tasks too hard, like setting environment variables, etc. And more advanced things like scripting, virtualisation, containers, etc. are generally easier to do in Linux.

E.g. a very important detail IMHO is the active vs. inactive modelines. In your light theme they are virtually the same, so you need to chase the cursor to find the active window. The first thing I modify in themes I use is to make modeline colours such that inactive one is faded but still legible, and active modeline really stands out: https://github.com/cadadr/configuration/blob/bf8b87c36dbab85d1ec35f3c9aa6f7d3c5e1f347/emacs.d/init.el#L5842 In general if you're not limited by a palette it's easier to adjust everything perfectly.

Selectrum's wiki has some basic info on completing-read: https://github.com/radian-software/selectrum/wiki/Tips-for-Creating-Commands

[1] https://github.com/radian-software/selectrum/issues/114

I noticed over the weekend that the venerable Selectrum package made by Radon Rosborough has been deprecated in favor of Daniel Mendler's Vertico package.

https://github.com/radian-software/selectrum is getting superceded by https://github.com/minad/vertico within just a year or two.

I have been using straight for a while now and I think it is great! The ability to lazy load everything by default does a lot to make Emacs snappier (or at the very least, faster to boot). Being able to pull packages directly from git (be it local or a forge) makes package development a lot easier. raxod has a lot of really sleek, modern emacs packages that I would encourage everyone to check out, spectrum[0] and ctrf[1] in particular are really great as well.

[0] https://github.com/radian-software/selectrum

selectrum

;;https://github.com/raxod502/selectrum (defun force-debug (func &rest args) (condition-case e (apply func args) ((debug error) (signal (car e) (cdr e))))) (advice-add #'selectrum--minibuffer-post-command-hook :around #'force-debug) (selectrum-mode +1) ;; to make sorting and filtering more intelligent (selectrum-prescient-mode +1) ;; to save your command history on disk, so the sorting gets more ;; intelligent over time (prescient-persist-mode +1) (setq completion-styles '(orderless)) ;; Persist history over Emacs restarts (savehist-mode) ;; Optional performance optimization ;; by highlighting only the visible candidates. (setq orderless-skip-highlighting (lambda () selectrum-is-active)) (setq selectrum-highlight-candidates-function #'orderless-highlight-matches) (setq selectrum-prescient-enable-filtering nil) (selectrum-prescient-mode +1) (prescient-persist-mode +1) (use-package marginalia :ensure t :config (marginalia-mode)) (use-package embark :ensure t :bind (("C-." . embark-act) ;; pick some comfortable binding ("C-;" . embark-dwim) ;; good alternative: M-. ("C-h B" . embark-bindings)) ;; alternative for \describe-bindings' :init ;; Optionally replace the key help with a completing-read interface (setq prefix-help-command #'embark-prefix-help-command) :config ;; Hide the mode line of the Embark live/completions buffers (add-to-list 'display-buffer-alist '("\`\Embark Collect \(Live\|Completions\)\" nil (window-parameters (mode-line-format . none))))) ;; Consult users will also want the embark-consult package. (use-package embark-consult :ensure t :after (embark consult) :demand t ; only necessary if you have the hook below ;; if you want to have consult previews as you move around an ;; auto-updating embark collect buffer :hook (embark-collect-mode . consult-preview-at-point-mode))`

Selectrum and Prescient would be nice.