Bram / is on Nostr

Tooting mostly about #emacs.

Indexed by tootfinder.

Last Notes

2024-08-03 20:22:20

New #Emacs package: elfeed-export, which exports your elfeed database to JSON or Emacs Lisp Data. I wrote it as a (readable) backup and to allow other tools to process the news entries in my feeds.
Still under development and refinement, and a bit underdocumented.
When you're interested, clone a copy of the package at:
https://codeberg.org/bram85/elfeed-export
#elfeed

2024-08-03 06:51:39

#kagi.el 0.6 is on MELPA Stable with full support for Markdown formatting in #fastgpt responses (using the markdown-mode package)
Another addition to kagi-fastgpt-shell is a welcome message when starting the shell.
https://codeberg.org/bram85/kagi.el
#emacs
https://media.emacs.ch/media_attachments/files/112/896/745/773/649/060/original/2f468075169d6972.jpg

2024-07-02 17:48:48

I made a new #emacs packa.. uhm, hack to remove #elfeed entries from its database. Use at your own risk.
By default it will remove old entries, but this package allows you to set your own predicates. This hack comes with an actual README with examples on how to configure it.
If you feel lucky clone the repo at:
https://codeberg.org/bram85/elfeed-prune

2024-05-01 11:52:57

`magit-find-file` allows you to open a file at a certain revision, which could be a (relative) timestamp. Notation: HEAD@{3 days ago}.
Using these bits, I wrote a small #emacs command that displays a file at a certain moment, chosen with the built-in calendar.
```
(defun bram85-magit-find-file-as-of (datetime)
(interactive (list (org-read-date)))
(let ((rev (format "HEAD@{%s}" datetime)))
(magit-find-file rev (magit-read-file-from-rev rev "File: "))))
```
#magit #git

2024-04-15 19:23:58

Now it's possible to access #kagi FastGPT from #orgmode:
#+begin_src kagi-fastgpt
A prompt inside an Org Babel block.
#+end_src
Check the README how to configure.
http://codeberg.org/bram85/kagi.el
#emacs
https://media.emacs.ch/media_attachments/files/112/276/844/500/562/823/original/e2f1232642e11e38.jpg

2024-02-09 19:18:13

Here's a function to generate QR codes inside #Emacs. Select a region or enter the text in the minibuffer and it will be encoded with the external `qrencode` tool. By default it generates an UTF8 QR code, with a universal prefix an PNG is created.
https://apps.bram85.nl/git/bram/gists/src/commit/73a87ed7675e9f71de3a7cb0cd2c43727dd34e05/gists/generate-qr.el
This is different from the qrencode.el package which is a native #elisp QR encoder:
https://github.com/ruediger/qrencode-el
https://media.emacs.ch/media_attachments/files/111/903/091/744/710/331/original/08e1a32edc77d891.jpeg

2024-01-26 18:19:54

kagi.el now has the `kagi-translate` command, a versatile command to translate text to another language using #Kagi #FastGPT
With `kagi-translate` you can:
- translate simple words from the minibuffer
- translate the region
- translate a whole buffer
- use it as a function to translate text from #elisp
- Not limited to spoken languages, you can convert a code buffer to another programming language (see demo below).
https://asciinema.org/a/633592
Should come to MELPA soon on the next build, or if you can't wait clone from the git repo at:
https://codeberg.org/bram85/kagi.el
As with all other kagi.el functionality, please check the project's README on how to get your API token, if you haven't done so already.
#emacs
https://media.emacs.ch/media_attachments/files/111/823/595/025/583/530/original/942db0aa4428fb76.mp4

2024-01-08 06:16:21

kagi.el is now available on MELPA and MELPA Stable.
https://melpa.org/#/kagi
This #emacs package provides access to #kagi's #FastGPT (inspired my chatgpt-shell) and the Universal Summarizer to summarize buffers, regions or URLs.
In the future I hope to implement the other APIs as well, which are more search oriented.
Check the README for the setup instructions.
https://codeberg.org/bram85/kagi.el/src/branch/main/README.org

2024-01-03 20:08:04

A while ago I shared a [toot] to create a denote entry from a URL. It fetches the URL
from the clipboard, retrieves the default title from the web and proposes it as such.
In the meantime I made this a bit smoother. Next to the clipboard, it will also look
at the text at point and the current elfeed entry to get URL candidates.
I put a short demo on Asciinema: https://asciinema.org/a/oWH0taGbOHbfOBcmdOsEO9wUu
The current implementation can be found on my site: https://apps.bram85.nl/git/bram/gists/src/commit/b04e6762031db22ea53de64c5e478fb16191\
e1f1/gists/denote-create-note-url.el
Previous toot: https://emacs.ch/@bram85/111150204263888493

2023-12-27 21:43:28

Added a section to the README to integrate my #kagi.el #emacs package with #embark.
This allows you to:
- Summarize buffer content from the minibuffer (e.g. from `switch-buffer`)
- Summarize a URL when calling `embark-act` while point is on a URL.
- Summarize the region when calling `embark-act`.
In all these contexts, hit 'K s' from the embark-act menu.
Might be turned as part of the package in some way.
https://codeberg.org/bram85/kagi.el#user-content-headline-7
https://media.emacs.ch/media_attachments/files/111/654/554/134/493/872/original/6a83ca23786fd8cb.mp4

2023-12-22 09:04:00

The screenshot in yesterday's toot on #fastgpt displays the question: what makes it better than ChatGPT?
The funny thing is it missed out on two major aspects:
1. The model is open source:
https://github.com/labring/FastGPT/
2. Also I like the fact more that Kagi provides the search results on which the answer was based. This gives more transparency than ChatGPT does.
Downside: it's not really a chat, it answers one question at a time without followup. But that is being worked on, apparently.
If you're curious, you can see for yourself at https://kagi.com/fastgpt.
https://emacs.ch/@bram85/111620681768693603

2023-12-21 22:07:18

Started a new #emacs package providing a shell for #kagi #FastGPT: kagi.el.
Of course, inspired by @npub1fh8…hm0p's fantastic chatgpt-shell package. In fact, both use the underlying shell-maker package for the heavy lifting.
Still very rough, lots of refinements to be done. But the basics work.
https://codeberg.org/bram85/kagi.el
https://media.emacs.ch/media_attachments/files/111/620/668/358/950/434/original/79baedbd038c8bdc.jpg

2023-08-30 19:52:17

Here's a function using #magit to show the diff of the current buffer since a certain date/time.
My use-case: I put my work notes in a single org file. When our daily standup starts I can quickly review what I worked on / wrote down in the last 24 hours.
https://apps.bram85.nl/git/bram/gists/src/commit/cb4c371e5ce68ba89afce0ae020e6b6f4161d237/gists/magit-diff-range-current-buffer.el
#emacs

2023-08-29 06:45:45

- reply

@npub1fzx…t6h3 Overall speaking I like the toki pona language, although I still have some doubts with the level of expression you can reach with the limited vocabulary. Time will tell.
A few observations after a few weeks:
- The limited vocabulary is a feature: it makes you more mindful about what you say and how to decompose something in more elemental concepts. I think that was also the intent of the language's creator.
- The grammar is simple and (almost) free of exceptions.
- I like the blend of many 'real' languages coming together: English, Dutch / German, Russian, Mandarin and more (but those are the ones I'm familiar with to different extents).
So at least it's a fun language to learn, its usefulness is yet to be seen. But learning a language is more than learning some words and rules, I already notice how it changes the way you think.

2023-07-02 19:41:16

#emacs package persist-state 0.3 available on (stable) MELPA.
This package regularly saves your Emacs state (bookmarks, history,
recent files, etc.) when idle for a brief moment. In case Emacs is
idle for a longer time, no state is saved.
Version 0.3 adds support for regularly saving your buffer positions using the builtin saveplace package.
https://stable.melpa.org/#/persist-state

2023-05-10 20:08:41

There's a Reddit thread on diagramming tools that works with #orgmode. I noticed that #ditaa is a drawing tool that still works, so I contributed to the thread:
https://www.reddit.com/r/emacs/comments/13d38zq/comment/jjn96u6/
Quoting myself:
ditaa still works within org-mode. The nice thing is that `artist-mode` kind of produces ditaa compatible ASCII drawings.
The (outdated) documentation states that the ditaa jar comes with Emacs, but that's no longer the case. So download it and then `setq org-ditaa-jar-path /path/to/ditaa.jar` in your config.
https://orgmode.org/worg//org-contrib/babel/languages/ob-doc-ditaa.html
https://ditaa.sourceforge.net/