Here is a small post about deploying a Matrix server in the homelab.

The goal is a Matrix server running on a FreeBSD jail that can be used from the web browser on your desktop.

It is a quick-and-dirty configuration, to get you started.

This setup is for testing purposes. The server will not be exposed to the internet and it will not federate with the Matrix network.

The setup contains three parts:

• the Conduit Matrix server
• the Cinny web Matrix client
• NGINX as reverse proxy

Domain names

First, create in your local DNS server two subdomain names, one for Conduit and one for Cinny, f.e., matrix.example.com and cinny.example.com, both pointing to the ip-address of the jail.

Certificates

Create certificates for the two domains.

If you haven't setup a local CA, do that first, for example using mkcert.

Place the certificate files in /etc/ssl/certs/ in the jail.

Conduit

Conduit advertises itself as a simple, fast and reliable matrix server.

It is available as a FreeBSD package.

Copy the file /usr/local/etc/conduit.toml.sample to /usr/local/etc/conduit.toml and edit this file:

Servername

This is the subdomain of your Conduitserver.

servername = "matrix.example.com"

Registration token

New users can register with this token.

registration_token = "very-secret"

Allow federation

We won't expose our server to the internet. Federation will not be required and we will not allow it.

allow_federation = false

Cinny

Cinny is yet another Matrix client. It runs as a web application. You access it with your web browser.

It is available as a FreeBSD package.

The package manager installs Cinny in /usr/local/www/cinny.

In this directory, copy the file config.json.example to config.json and edit it:

homeserverList

Delete all entries in the homeserverList and insert the subdomain of your Conduit server.

"homeserverList": [
  "matrix.example.com"
],

featuredCommunities

Probably you can delete a lot here. I only cleaned the list "servers":

"servers": ["matrix.example.com"]

NGINX

NGINX is a very great web server.

It is available as a FreeBSD package.

Here we use it as a reverse proxy.

In the file /usr/local/etc/nginx/nginx.conf add two blocks for the virtual servers:

server {
    listen *:443 ssl;
    server_name matrix.example.com;
    client_max_body_size 1G;
    ssl_certificate /etc/ssl/certs/matrix.example.com.crt;
    ssl_certificate_key /etc/ssl/certs/matrix.example.com.key;
    location / {
       proxy_pass http://127.0.0.1:6167;
    }
}

server {
    listen *:443 ssl;
    server_name cinny.example.com;
    client_max_body_size 1G;
    ssl_certificate /etc/ssl/certs/cindy.example.com.crt;
    ssl_certificate_key /etc/ssl/certs/cindy.example.com.key;
    location / {
       root /usr/local/www/cinny;
       index index.html;
    }
}

Allow the daemons to run

In /etc/rc.conf add two lines to allow the daemons to run.

conduit_enable="YES"
nginx_enable="YES"

Start or restart the Conduit and the NGINX daemons.

Create a few accounts

Open cinny.example.com in your web browser.

Register as new account. Choose your server (matrix.example.com), and provide a user name and a password. Insert the registration token from the Conduit configuration (above this is mentioned as "very-secret") in the field below the two password fields.

Repeat, to get a few accounts to play with.

Have fun playing with Matrix in you homelab!

Below are the two books that I finished this month. I started a third one, but didn't like and decided not to finish it.

Non-fiction

The Beak of the Finch: A Story of Evolution in Our Time by Jonathan Weiner (1994)

• 332 pages
• 5 stars

A book about scientific research aimed at better understanding natural selection, sexual selection, and evolution.

Much of the book is about two biologists, Peter and Rosemary Grant, doing a decades long research of finches on a small vulcanic island in the Galapagos. Also some other evolutionairy biological research is discussed. The book explains how natural selection and evolution can indeed be seen in action, contrairy the old idea that evolution moves too slow to witness.

I learned from it that in nature there is always competition and that in hard times, like when food is extra scarce, the selection pressure increases and evolution accelerates.

A very interesting book, easy to read and understand.

Fiction

A Maze of Death by Philip K. Dick (1970)

• 216 pages
• 5 stars

Science fiction story of a very small group of people colonizing the planet Delmak-O. They arrived in so-called "nosers", which are one-way space ships. The story develops when bad things starting to happen. This is, as often with works from this writer, a rather short book.

It is a good story. with enough tension and surprises to keep you engaged. Recommended!

Texinfo

Texinfo is a brilliant hypertext system that predates HTML. You access it with info, which is a text mode application. The content consists of a number of nodes. Each node is an information element. You access the nodes through menus, hyperlinks, and indices. Also you can "walk" through a info file with keys like [ and ]. See info info.

Leveraging the power of TeX, Texinfo files can be used to produce gorgeous looking PDF files, like the GNU books. Compared to generating PDF files from LaTeX, generating PDF files from Texinfo is incredible fast.

po4a

The po4a (PO for anything) eases documentation translations and their maintenance.

This is what this website says:

It extracts the translatable material from the original document, and places it into a PO file that is adapted to the translation process. Once this PO file is updated by the translator, po4a re-injects the translation into the structure of the original document to generate a translated document.

po4a supports several documentation formats, including Texinfo.

Workflow

In general, the workflow is something like this:

• Create a PO-file from the orginal Texinfo file.
• Add the translations to the PO-file.
• Generate a translated Texinfo file.
• Use the translated Texinfo file like any other Texinfo file, e.g., create an Info file and a PDF with it.

When the original Texinfo file changes:

• Update the PO-file.
• Add or change translations where needed.
• Generate a new translated Texinfo file.

Store the original Texinfo file

In order for the update of the PO-file to work, you'll need both the original Texinfo file as well as the new version of the Texinfo file. It is therefor important to keep a copy of the original Texinfo file.

In the examples below, two versions of the Texinfo file are mentioned:

• testinfo-v1.texi: this represents the original file to be translated.
• testinfo-v2.texi: this represents a new version of the file to be translated.

Create a PO-file

To create the PO-file from the original Texinfo file, using the following command.

po4a-updatepo \
--format texinfo \
--master testinfo-v1.texi \
--po testinfo.po

Replace testinfo-v1.texi with the name of the original Texinfo file.

Replace testinfo.po with the desired name of the PO-file.

Translating the PO-file

The PO-file contains a lot of strings and paragraphs to be translated.

Translating strings

The entry for a string in the PO-file is something like this:

#. type: menuentry
#: testinfo-v2.texi:30
msgid "Fundamental stuff"
msgstr ""

The first line containing the type differs for the roles of the string in the Texinfo document.

• The line starting with msgid is a line to be translated.
• The line starting with msgstr is where the translation must be paced.

Translating paragraphs

The entry for a paragraph looks something likes this:

#. type: Plain text
#: testinfo.texi:24
msgid ""
"In this chapter I define the concepts that will be used throughout the rest "
"of the document.  Moreover, measures of efficiencies as well as bounds of "
"complexity will be introduced."
msgstr ""

The translation of the paragraph must be placed right below the line starting with msgstr:

#. type: Plain text
#: testinfo.texi:24
msgid ""
"In this chapter I define the concepts that will be used throughout the rest "
"of the document.  Moreover, measures of efficiencies as well as bounds of "
"complexity will be introduced."
msgstr ""
"In dit hoofdstuk definieer ik de concepten die gebruikt worden in de rest "
" ... "

The lines in the translated paragraph start with a double quote (") and end with a space and a double quote.

The last line of the translated paragraph ends with a double quote without a space, just like the last line of the paragraph in the original language.

(Example text copied from Writing Documentation, Part IV: Texinfo on Linux Gazette.)

Texinfo commands in paragraphs

Texinfo commands require special formatting of the lines.

For this, we make use of the \n formatting in the strings and paragraphs:

String example:

#. type: smallexample
#: testinfo.texi:1421
#, no-wrap
msgid "Original line to be translated\n"
msgstr "Translated line\n"

Sometimes po4a has recognized the Texinfo command and treated as such, which can be seen at the type of the PO-element, as here above.

Other times the Texinfo command appears inside a paragraphs to be translated. Here we make uses of careful \n formatting in the translated paragraph:

" ... "
"Some text inside the paragraph\n"
"@ifnottex \n@xref{Some-ref}.\n@end ifnottex\n"
"@iftex \n@xref{Some-other-ref}. \n@end iftex\n"
"Some other text inside the paragraph. "
" ... "

Not: the \n is directly followed by a double quote, without a space before the double quote.

Update a PO-file

When later the original Texinfo file gets an update, we have to reflect those changes in our translated file.

Here's how to update the PO-file. Note the use of the name of the new version of the Texinfo file (testinfo-v2.texi) and the name of original file the PO-file was created with (testinfo-v1.texi).

po4a-updatepo \
  --format texinfo \
  --master testinfo-v2.texi \
  --previous testinfo-v1.texi \
  --po testinfo.po

The PO-file in the command above is the same file we created in the first step and has been filled with our translations.

How to recognize changes in the updated PO-file

After running the po4a-updatepo command, the PO-file with our translations is changed. po4a has merged the changes into it.

This means we must go through the PO-file again, to add or change translations. Below follows how to recognize new text, deleted text and changed text.

New text

New lines and paragraphs can be recognized by an empty msgstr. Example:

msgid "Very new chapter"
msgstr ""

Deleted text

Deleted text can be recognized by a pound char (#) at the start of the msgid and msgstr lines. Example:

#~ msgid "This line will be deleted in the next version."
#~ msgstr "Deze regel wordt in de volgende versie verwijderd."

Changed lines

Changed lines are preceded by a line with #| msgid at the start. Example:

#| msgid "This line will be changed in the next version."
msgid "This line is now changed in this version."
msgstr "Deze regel wordt in de nieuwe vesie gewijzigd."

• The line starting with #| msgid is the line as it was in the previous version of the Texinfo file.
• The line starting with msgid is the line as it is in the new version of the Texinfo file.
• The msgstr contains the previous translation, is no longer valid and must be changed into a translation of the new text.

Create translated texinfo file

po4a-translate \
  --format texinfo \
  --master testinfo-v2.texi \
  --po testinfo.po \
  --localized testinfo-nl.texi

Here, the input file testinfo-v2.texi is used as an example. The output file is testinfo-nl.texi, the Texinfo file that is generated using the PO-file and the Texinfo file it was created from.

For convenience, create a Makefile for this:

testinfo-nl.texi:   testinfo.po
    po4a-translate \
      --format texinfo \
      --master testinfo-v2.texi \
      --po testinfo.po \
      --localized testinfo-nl.texi

When only a part of the translation has been done, po4a will refuse to generate the Texinfo output file. This can be configured by using the switch -k, e.g., -k 30 to generate an output file when at least 30% of the translation has been done. The default is 80%.

Generate a PDF file from the texinfo file with the translation

makeinfo --pdf testinfo-nl.texi

To add this to the Makefile, you could do something like this:

testinfo-nl.pdf:    testinfo-nl.texi testinfo.po
    makeinfo --pdf testinfo-nl.texi

testinfo-nl.texi:   testinfo-nl.po
    po4a-translate \
     --format texinfo \
     --master testinfo.texi \
     --po testinfo.po \
     --localized testinfo-nl.texi

When there are some non fatal errors, the makeinfo command can be extended with the switch ---force to generate an output file even when there are errors.

Editing PO-files with Emacs

Here follow some tips for working in Emacs.

Spell check

When working on the PO file and inserting translations, the file will contain text in two languages, the original language and the language we are translating to.

An easy way to spellcheck a single entry is by marking it as a region and run M-x flyspell-region.

First, set the dictionary to be used with M-x ispell-change-dictionary.

Flyspell-mode

Flyspell-mode checks your spelling while you type. Just enable it --after having changed to the right dictionary-- with: ````

M-x flyspell-mode

Curly braces

Texinfo uses a lot of curly braces. Make live easier and enable electric-pair-mode:

M-x electric-pair-mode

This will let Emacs insert the closing curly brace for you.

.dir-locals.el

Set several options automagically with a file named .dir-locals.el in your directory, for example:

((nil . ((ispell-local-dictionary . "nl")
     (abbrev-mode . t)
     (eval . (flyspell-mode))
     (eval . (electric-pair-mode))
     (eval . (company-mode)))))

Where nl is the Dutch dictionary.

View local info file in Info-mode

To open in a local info file in info-mode, use C-u C-h i.

Working with paragraphs

For me, this works convenient:

• First, write the paragraph with the translation just as normal text.
• Next, when needed, correct the paragraph using fill-paragraph, normally with M-q.
• Third, surround each line with double quotes, with a space before the double at the end of the line.
• Make sure the last line just ends with a double quote, not a space and double quote.

Some small Elisp functions can help:

(defun my/quote-region ()
  "Surround each line in the region with quotes."
  (interactive)
  (save-excursion
    (save-restriction
      (narrow-to-region (region-beginning) (region-end))
      (goto-char (point-min))
      (replace-regexp "^" "\"")
      (goto-char (point-min))
      (replace-regexp "$" " \""))))

(defun my/unquote-region ()
  "Remove quotes around the lines in the region."
  (interactive)
  (save-excursion
    (save-restriction
      (narrow-to-region (region-beginning) (region-end))
      (goto-char (point-min))
      (replace-regexp "^\"" "")
      (goto-char (point-min))
      (replace-regexp "\"$" ""))))

(defun my/unquote-region-and-fill-paragraph ()
  "Remove quotes around the lines in the region and fill-paragraph"
  (interactive)
  (save-excursion
    (save-restriction
      (narrow-to-region (region-beginning) (region-end))
      (goto-char (point-min))
      (replace-regexp "^\"" "")
      (goto-char (point-min))
      (replace-regexp "\"$" "")
      (goto-char (point-min))
      (fill-paragraph))))

Example key bindings:
(define-key global-map (kbd "C-c n q") #'my/quote-region)
(define-key global-map (kbd "C-c n Q") #'my/unquote-region)
(define-key global-map (kbd "C-c n m") #'my/unquote-region-and-fill-paragraph)

The function my/quote-region can be improved: it leaves a space before the double quote on the last line.

I started this year with re-reading "Pattern Recognition". This is the first book that I re-read within a year.

Because I really enjoyed this book, I followed it with the other two books of the Blue Ant trilogy.

Non fiction

Ends of the Earth: Journeys to the Polar Regions in Search of Life, the Cosmos, and Our Future by Neil Shubin (2025)

• 288 pages
• 5 stars

A book about Antarctic and Arctic expeditions, science, and history.

The book opens with the author, at the time still a student, experiences his first polar expedition. It is the start of a career with 30 years of science and many more polar expeditions.

The book tells about survival on the poles and the life of scientists working in extreme environments, and couples that with historic events. Well written, this book explains how the poles influences our weather and climate, and all life on earth, and the dangers of the current climate crisis. It is easy to read, never boring.

Absolutely recommended reading.

Fiction

Pattern Recognition by William Gibson (2003)

• 368 pages
• 5 stars

A great and captivating story that develops at a good pace. Real and easy to identify with protagonist. Lovely written, with many great sentences, great descriptions of the surroundings, and of the characters. A real and highly recommended page-turner.

Re-read this book. Gibson creates wonderful sentences and I expect to have missed several in the first read, being captivated by the story development. I like Cayse, the protagonist and her way of thinking. This book is definitely becoming a kind of comfort read for me.

Spook Country by William Gibson (2007)

• 371 pages
• 5 stars

Second book in the Blue Ant trilogy. Like the first book, "Pattern Recognition", this is not a science fiction story. Although I felt less connection with the three protagonists, I got quickly immersed into the story. Again the book is wonderful written, good plot, and a page-turner.

Zero History by William Gibson (2010)

• 416 pages
• 5 stars

Third book in the Blue Ant trilogy. Again not a science fiction book. And again a story around the fashion world. I got quickly immersed into the story. Another page turner, wonderful written. I enjoyed this one more then the second book from the trilogy.

These are all three great books, Pattern Recognition is the best, followed by ero History, and Spook Country on third.

I started a new web page, to publish a collection of awesome and inspiring web pages that are worth reading.

Currently the collection is modest, I hope to build over time.

Find this collection of evergreens here.

Implement a simple review process for denote notes with denote-review

How an Emacs package came into existence

Systems in use

Over time I have grown into using several different note systems next to each other, like taking notes in awkiawki wiki and in denote.

Both were a succession of earlier systems:

• My notes in awkiawki once started as notes in vimwiki.
• My notes in denote once started as notes in Deft for Emacs.

OBTF

Thinking about the best way forward I experimented with different options.

Moving towards an OBTF --One Big Text File-- looked promising. I explored org mode options like org refile. To get familiarized I set up some exercises, configured an org capture for an OBTF and started using it.

After getting some hands on experience, I came to the conclusion that I prefer a lot of small note files above one big one.

Migrated awkiawki

I decided to migrate the awkiawki notes to denote and wrote a small Guile Scheme script to convert the wiki formatted notes into denote formatted notes (in org mode). Links between wiki-notes became links between denote notes.

After the migration I ended up with over 1,300 notes in denote.

Need for a reviewing process

All notes are created equal, but some are more equal than others.

It was clear that I needed a system to review the collection of notes, get rid of notes that addressed the same topic or were outdated, and which notes are underdeveloped. Knowing myself, this would result in a early death without a systematic approach.

I started thinking about a simple set up, that would not get in your way and wouldn't affect being "in the flow" when working with notes.

Denote notes

Notes are unstructured data by default, but denote notes start with some structured data, called "frontmatter". It contains metadata like the title, keywords (think: tags), an identifier and the creation date of a note.

This seemed the best place for a small addition. I thought about setting a deadline for each note, before which it had to be reviewed, but that opens another rabbit hole and I decided against it.

I ended up with creating a small system that added an extra field to the frontmatter, a line that stated the last review date, and a method to select some notes and order these by that review date.

I build this and brought it under the attention of Protesilaos Stavrou, the creator of denote, also known as "Prot". Denote supports different kind of formatting of the frontmatter, markdown-toml, markdown-yaml, plain text, and org mode. I had build my system for my own needs, and it just supported org-mode.

Prot nudged me into making the system more general, by adding support for the other formats. This turned out less complicated than I had originally envisioned and the system grew into a usable Emacs package. I submitted the package for inclusion in GNU ELPA. Philip Kaludercic and Stefan Monnier helped me improved the code, and the rest is history.

Usage

denote-review adds an extra line to the frontmatter. Here's an example of the frontmatter of an org-mode formatted denote note:

The function denote-review-set-date adds the reviewdate to the frontmatter, or updates it to the current date.

For convenience, bind this to a keychord, for example:

(define-key global-map (kbd "C-c n x") #'denote-review-set-date)

To get started, select a number of notes in Dired and bulk-insert the frontmatter into these notes, see below.

After extending the frontmatter of several notes with a review data it becomes useful to list them, using the command:

 M-x denote-review-display-list

This command prompts for a denote-directory when needed (when there is more than one denote-directory in your Emacs configuration, or when you have set up several directories using denote-silo). Select a directory using completion.

Next, the command prompts for a keyword to narrow the list to. Choose a keyword or just hit RET (the enter key) for all notes.

This creates a tabulated list with for each note the review date and the file name:

Only notes that contain the reviewdate frontmatter are displayed, notes without a review date are ignored.

The notes are sorted by review date. Click on one of the column headers to order by the contents of that column or to reverse the sort order. Using the key S (uppercase, so shift-s) performs the same actions.

Select a line by moving point up or down in this list. With a single key press, perform one of the following commands:

The tabulated list mode comes with a number of build-in commands, like p and n to move to the previous or next line, SPC and DEL (space bar and backspace key) to scroll down or up.

The general idea is that you from time to time set up a review sessions:

• select a number of notes, f.e., notes with a specific keyword
• go through the list and review notes, for each note edit where needed
• update the reviewdate of reviewed notes

Slowly moving to the situation where all notes have a recent review date.

Note: The filenames in the screenshots above are based on headers in the README from Protesilaos Stavrou, see protesilaos.com/emacs/denote

Bulk-insert the review date

Open a denote-directory in Dired and mark one or more notes with m.

Unmark a marked note with u.

To bulk-insert the review date there are two options:

The original creation date as review date

Issue the command

M-x denote-review-set-date-dired-marked-files

This will insert a frontmatter line with the review date derived from the date encoded in the first chars of the filename.

The current date as review date

Enter the Universal Argument C-u

Issue the command

M-x denote-review-set-date-dired-marked-files.

This will insert a frontmatter line with the current date a review date.

Select multiple notes in Dired

Like Prot has written in his denote manual, it is easy to select and mark a number of notes in Dired using a regexp.

For example, to select and mark all notes with the keyword "foo":

• press %m
• enter _foo RET

To reverse the selection (all notes that don't have the keyword "foo"):

• press %m
• enter _foo RET
• press t (toggle)

Happy note taking!

Project Gutenberg is a fantastic treasure trove filled to the brim with over 75,000 books.

In 2025 I read several books and short stories that are now in the public domain. Here are the most noteworthy.

Books

Little Fuzzy by H. Beam Piper (1962)

252 pages

Short science fiction novel about a planet in deep space which is being mined for "sunstones". The discovery of a small humanoid furry race set of a chain of events. This book breathes US culture, but is non the less a lovely read.

Gutenberg book 18137

Ziska - The Problem of a Wicked Soul by Marie Corelli (1898)

206 pages

A Gothic novel set in Egypt. A group of English tourists visits Cairo when the mysterious "Princess Zika" appears. Two friends fall in love with her. A spooky story about love and revenge.

Gutenberg book 5079

Short stories

Let's Get Together by Isaac Asimov (1957)

18 pages

Short story set in a world where the Cold War has endured for a century.

Gutenberg book 68377

The Black Kiss by Robert Bloch, Henry Kuttner (1937)

20 pages

A horror short story. I almost never read horror stories, this one I enjoyed.

Gutenberg book 76435

In the heyday of blogs, many blogs published a blogroll, often is a side bar.

This was a list with which showed a selection of links other blogs. It showed the blogs that had recently been updated.

I created a blogroll that does the same.

Selection of links

A script looks every night at the personal websites and blogs listed on my links page. When a change in their RSS-feed is detected, the website or blog is listed in the blogroll.

The demise of the search engines and the embrace of untrustworthy, useless AI-generated results, optimistically referred to as "AI-powered" are upon us.

The web is turning into a web for profit. Individual websites and blogs, that make up the smol web, are almost impossible to find.

Some websites are trying to do something about this, by either creating some kind of feed aggregator, or create a search engine that targets small websites, often based on the parsing of RSS feeds.

New category: Feed aggregators

In my links I have now added a new category, "Feed aggregator" for this.

Currently there are only a few entries in it, but I hope to grow this list over time.

Let's take back the web!

The Generation Lissa! webring is for people who learned how to build their own personal websites with code between the years of 1996 and 2005, and still run a personal space on the Internet.

More about the webring

Webrings help to keep the web alive

Small hand-cafted, non commercial websites are almost impossible to discover.

The web gets demolished by the demise of the search engines and the embrace of untrustworthy, useless AI-generated results, optimistically referred to as "AI-powered".

Search engines prioritize large domains with commercial content. The web is swamped with "SEO-optimized" content and AI-generated slob.

Personal blogs, art pages, and other hand-crafted websites practically never turn up in search results.

The best way to fight this, is to restore the wanderability of the early web: Create links between websites and make following these links fun.

Embrace webrings

The blogosphere made the web a place of personal expression. The current craze is massive delivery of shallow, meaningless content.

Webrings bring back the joy of surfing. They encourage you to explore, and frequently lead you to unexpected discoveries. The visitors drift through a community instead of being funneled by billionaire’s dark algorithms.

The open web needs you

Website creators require motivation and inspiration. Without visitors, homemade websites may go neglected.

• If you have a website, join a webring. Or maybe more than one.
• When visiting a website, follow the webring links, and see what happens.

Local XMPP server on a Raspberry Pi Zero 2 W

This is an article about a local server only, not reachable from the internet.

The differences with a XMPP server that is connected to the internet and does federation with other XMPP servers is in the infrastructure: TLS-certificates from an official CA, and official DNS records. When providing a XMPP server on the internet, security and privacy protection are very important.

Light weight server

Prosody is a light weight, easy to administrate XMPP server. I had it running for years on a BeagleBone Black.

Due to circumstances I had to move that XMPP server to a location outside of our home network.

I still wanted a local XMPP server just for testing purposes and set up a Prosody server on a small Raspberry Pi Zero 2 W running Raspberry Pi OS Lite.

This page describes the use of a Raspberry Pi as server, but it is not limited to that platform. Prosody is available for all the open source operating systems.

The use of a local only XMPP server, not federated and not reachable from the internet is of course limited. It can be used for test purposes, local machine-to-machine communication, and it is fun!

The main objective is a server with the following capabilities:

• support for OMEMO encryption
• support for Multi-User Chat
• support for HTTP File Upload

Requirements

DNS

Three subdomains are needed:

• a subdomain for the XMPP server
• a subdomain for the Multi-User Chat
• a subdomain for the HTTP-File-Upload

All three subdomains can point to the same IP address, the address of the Raspberry Pi. Set it up for both the ipv4 and the ipv6 address. (Maybe only ipv6 works too, I have not tested that.)

The best solution is to set up a local DNS server in your local network.

If you don't have a local DNS server in your home lab, stop now and set one up first. It won't take much effort and is great for now and in the future. There a several open source servers available.

Unbound is a good DNS server, easy to set up, also for reverse look-ups.

Certificates

Each of the three subdomains needs a TLS certificate. In your home lab there is of course no need for certificates from an official CA.

The best solution is to set up a local CA.

When you run a local CA, you yourself can issue signed certificates for all the subdomains in your home lab.

Mkcert is an easy way to set up a CA and issue certificates. It can also be done from the command line with openssl.

Create and sign certificates for the three subdomains. Six files in total, two for each subdomain. Place the certificate files in /etc/prosody/certs/ or in /usr/local/etc/prosody/certs/, depending on your operating system.

Install and configure Prosody

Just use the package manager to install Prosody. It is available as package on most systems.

The location of the configuration files and work directory depends on your operating system. The configuration file prosody.cfg.lua will most likely be in /etc/prosody or in /usr/local/etc/prosody.

On the Raspberry Pi, the work directory is in /var/lib/.

Here is a complete configuration, for use in a home lab.

admins = { }
modules_enabled = {
        "disco"; -- Service discovery
        "roster"; -- Allow users to have a roster.
        "saslauth"; -- Authentication for clients and servers.
        "tls"; -- Add support for secure TLS on c2s/s2s connections
        "blocklist"; -- Allow users to block communications with other users
        "bookmarks"; -- Synchronise the list of open rooms between clients
        "carbons"; -- Keep multiple online clients in sync
        "dialback"; -- Support for verifying remote servers using DNS
        "limits"; -- Enable bandwidth limiting for XMPP connections
        "pep"; -- Allow users to store public and private data in their account
        "private"; -- Legacy account storage mechanism (XEP-0049)
        "smacks"; -- Stream management and resumption (XEP-0198)
        "vcard4"; -- User profiles (stored in PEP)
        "vcard_legacy"; -- Conversion between legacy vCard and PEP Avatar, vcard
        "csi_simple"; -- Simple but effective traffic optimizations for mobile devices
        "invites"; -- Create and manage invites
        "invites_adhoc"; -- Allow admins/users to create invitations via their client
        "invites_register"; -- Allows invited users to create accounts
        "ping"; -- Replies to XMPP pings with pongs
        "register"; -- Allow users to register on this server using a client and change passwords
        "time"; -- Let others know the time here on this server
        "uptime"; -- Report how long server has been running
        "version"; -- Replies to server version requests
        "mam"; -- Store recent messages to allow multi-device synchronization
        "admin_adhoc"; -- Allows administration via an XMPP client that supports ad-hoc commands
        "admin_shell"; -- Allow secure administration via 'prosodyctl shell'
        "posix"; -- POSIX functionality, sends server to background, enables syslog, etc.
        "groups"; -- Shared roster support
        "http";    -- Allow Prosody modules to expose various services over HTTP
        "http_file_share"; -- Let users share files via HTTP
}
pidfile = "/run/prosody/prosody.pid";
c2s_require_encryption = true
s2s_secure_auth = true
limits = {
    c2s = {
        rate = "10kb/s";
    };
    s2sin = {
        rate = "30kb/s";
    };
}
authentication = "internal_hashed"
archive_expires_after = "1w" -- Remove archived messages after 1 week
log = {
    debug = "/var/log/prosody/prosody.log";
    error = "/var/log/prosody/prosody.err";
    { levels = { "error" }; to = "syslog";  };
}
certificates = "certs"
VirtualHost "xmpp.your-domain.home"
    disco_items = {
        { "upload.your-domain.home", "file sharing service" },
        { "muc.your-domain.home",  "chat room service" },
    }
    ssl = {
        key = "/etc/prosody/certs/xmpp.your-domain.home.key";
        certificate = "/etc/prosody/certs/xmpp.your-domain.home.crt";
    }
    Component "upload.your-domain.home" "http_file_share"
        http_file_share_expire_after = 86400*28 -- 28 days
        http_file_share_size_limit = 128*1024*1024+16 -- 128 MB
    Component "muc.your-domain.home" "muc" name = "chat room service"
        modules_enabled = { "muc", "muc_mam", "vcard_muc","pastebin" }
        restrict_room_creation = false

Here we use three domains:

• xmpp.your-domain.home
• muc.your-domain.home
• upload.your-domain.home

Replace your-domain.home with the domain used in your home lab.

Make sure the line with the pidfile fits your operating system.

You can choose any name for the subdomains xmpp, muc, and upload. Be sure the names are the same as set up in the DNS server.

Make sure that the element disco_items comes before the Component-sections in your config. The order is important.

In this example, the logging is configured at the debug level. When everything is running smoothly, you can replace that with info.

Check your settings with prosodyctl check.

Start the service

Depending on your system you need to give some commands to get the server running.

Firstly, enable the server with systemctl enable prosody on systems with systemd, or f.e., add a line to /etc/rc.conf or /etc/rc.conf.local.

Next, start the server with systemctl start prosody or /etc/rc.d/prosody start or /usr/local/etc/rc.d/prosody start.

Check the service with prosodyctl status and check the log files.

Add a user

Add a user, in this example "alice", with:

prosodyctl adduser alice@xmpp.your-domain.home

This will prompt for a password and for a confirmation of the password.

Prepare for testing

Make sure the certificates are trusted on your desktop machine.

Install a XMPP client. Make sure it is ready for OMEMO and test that you can log in.

Have fun!

OpenBSD comes with the vmm(4) hypervisor and vmd(8) daemon. It is easy to create and install an OpenBSD virtual machine on an OpenBSD workstation. For more information, see the Virtualization page in the OpenBSD FAQ.

Running one or more OpenBSD virtual machines on your OpenBSD laptop is great for testing purposes.

VNC to local vm

A graphical desktop can be useful when testing or trying out stuff. Here we are going to set up VNC on a virtual machine that gives us an XDM login window.

The virtual machine we used is running OpenBSD 7.8.

XDM login over VNC on a OpenBSD vm

Install TigerVNC

We need to install Tiger VNC on the VM for this.

doas pkg_add tigervnc

Create a VNC passwordfile

We are going to need a password file on the VM for the VNC connection.

doas vncpasswd /etc/X11/xenodm/VNCpwd

Here, /etc/X11/xenodm/VNCpwd is the file name for the file we are going to create. Choose a path and file name to your liking.

The command vncpasswd will ask for a password and a confirmation of it.

Optionally you can also add a second password, for a read-only VNC session. We skipped this.

Edit xenodm-config

On the VM, edit the file /etc/X11/xenodm/xenodm-config. At the bottom, add a line:

DisplayManager._1.authorize: false

Edit Xservers

Edit the file /etc/X11/xenodm/Xservers.

At the bottom, comment out the line:

:0 local /usr/X11R6/bin/X :0 vt05

And add a line:

:1 local /usr/local/bin/Xvnc :1 -geometry 1024x768 -depth 16 -desktop vm1 -Passwordfile /etc/X11/xenodm/VNCpwd

Make sure the path and file name of the password file correspondent with the password file you just created.

Enable and start xenodm

On the VM, enable and start xenodm.

doas rcctl enable xenodm
doas rcctl start xenodm

Or, if xenodm is already running, restart the service.

Connect with vncviewwer

To get vncviewer on your laptop, install tigervnc.

doas pkg_add tigervnc

Now, connect to the VNC server on the VM:

vncviewer 100.64.1.3:1

Replace 100.64.1.3 with the ip address of your VM.

Optionally encrypt traffic

Remember that the VNC protocol transports your data unecrypted. This is insecure.

We use this setup only for testing purposes, only for local VM's on our laptop, without any sensitive data..

To protect your data from eavesdropping, encrypt the connection, f.e., with an SSH-tunnel.

Virtual machines are very useful. Creating and installing an OpenBSD virtual machine on an OpenBSD laptop is easy.

Have fun!

9front

9front (or plan9front) is a fork of the Plan 9 from Bell Labs operating system.

9front with Mothra and Netsurf

vm-bhyve

FreeBSD vm-bhyve is a great tool to create, install and manage virtual machines.

vm-bhyve is designed to work with ZFS, which makes it very easy to create new vm's, and make ZFS snapshots of those.

Virtual machines can either run with or without a graphical desktop. For machines running without a graphical desktop we use SSH.

The FreeBSD bhyve hypervisor provides VNC access to the virtual machines that boot with UEFI. Expose the VNC port to the network to allow for remote management.

9front on bhyve using vm-bhyve

This is how I got 9front installed and booting after the install.

I have exactly zero experience with 9front or plan9. so maybe there are better ways.

Download install iso

An install iso can be downloaded from 9front.org.

I downloaded 9front-11321.amd64.iso.

Installation

We don't use the normal way to install an operating system on a VM with vm-bhyve (vm install vm-name iso-name). This is because vm-bhyve removes the CD-ROM from the VM at the first reboot after this command.

We follow a slightly different way:

Create a new virtual machine:

vm create -t dragonfly 9frontvm

We choose the dragonfly template because is comes with UEFI and VNC support.

Add lines for the CD-ROM

Add the following lines to the configuration of the VM (the file /bhyve/9frontvm/9frontvm.conf):

disk1_type="ahci-cd"
disk1_dev="custom"
disk1_name="/bhyve/.iso/9front-11321.amd64.iso"

Start the installation:

vm start 9frontvm

Now open your VNC viewer to the VM.

From here you can follow the installation guide. The installation window doesn't show a cursor. just trust that you are indeed typing at the prompt.

Just accept the default values.

• At the bootargs just hit Enter.
• After selecting 1024x768x16 for the vgasize (just hit Enter) I had to select vga as monitor.

After accepting the default for the mouse port a graphical desktop is started.

In the terminal, issue the command inst/start.

Again, accept the default values.

• Disk to partition: sdE0.
• Install mbr or gpt: gpt.
• Use esp as 9fat: yes.

At the configuration of the network I did choose manual. This gives the option to enter ip address, netmask, gateway and DNS.

• Distribution disk: /dev/sdF0/data.

The list with options for the timezone for Europe is smaller than usual.

Time Zone: CET.

After the reboot, use the bootargs local!/dev/sdE0/fscache.

Prepare for easy boot

Next, we are going to edit 9plan.ini in the 9front terminal.

9fs esp
cd /n/esp
ed plan9.ini

There are more options beside ed to edit the file, f.e., use the acme editor.

Change the bootargs line to:

nobootprompt=local!/dev/sdE0/fscache

And add a line:

user=glenda

Save the changes, and check with cat plan9.ini.

If all is well, reboot (see below).

Now it will boot without the need to hit Enter and it will directly go into Rio (the graphical desktop).

Shutdown

Shut down the virtual machine by issuing the command fshalt in a 9front terminal, or use fshalt -r to reboot..

Desktop resolution

As 1024x768 is quite cramped, I changed the resolution to 1200x720.

First, I changed the line in plan9.ini:

vgasize=1200x720x16

Next, I added a line to the configuration of the VM:

graphics_res="1200x720"

Remote access using drawterm

VNC on bhyve is not the best experience. The mouse often requires some persuation to before reachting the edges of the frame, and it is not fast.

Drawterm solves this all.

I followed this turorial to allow for remote access using drawterm.

To get the latest version, get drawterm from drawterm.9front.org.

Resources

• wiki.9front.org/freebsd-bhyve
• Plan 9 desktop guide
• Setting up 9front an a Raspberry Pi with useful information, not only for Raspberry Pi.

Thanks

Thanks to Tom for the provided help, inspiration and useful links!

Have fun with 9front!

Updates

• 2025-01-01: added section on remote access with drawterm.

Running OS X Mavericks on an old mid 2013 MacBook Air

Specs of the Macbook Air

• 1.3 GHz Intel Core i5
• 4 Gb 1600 MHz DDR memory
• Intel HD Graphincs 5000
• 13 inch TFT LED 1440x900 display
• 128 GB SSD

Apart from the Apple keyboard layout, the keyboard is fine.

The glass trackpad (without buttons) is quite large. Unfortunately, Apple keyboards don't come with a TrackPoint like ThinkPads do, but the glass trackpad will do fine.

OS X Mavericks

Sometime ago I did a factory reset on this old MacBook. This installs the default operating system, which is OS X Mavericks. The old MacBook Air works fine with it. The laptop doesn't feel old, everything is snappy and the display is a joy to look at.

I decided not to upgrade to a newer version of OS X, or replace it with Linux or one of the BSDs. I like the look and feel of OS X Mavericks, it doesn't require an iCloud account to run, and it is completely different from my daily drivers: ThinkPads running FreeBSD with the ratpoison window manager, and FreeBSD with the Emacs X Window Manager "EXWM".

Of course running an old system has it challenges.

Here follows some information about the software I installed, and other stuff.

MacPorts

It's hardly believable, MacPorts is available for OS X Mavericks!

Using MacPorts I installed the following packages:

• Emacs 30.2 under X11
• GnuPG 2.4.8
• Hunspell
• Links
• Lynx
• OpenSSH
• OpenSSL
• RCS
• Tmux
• Xorg

See also my page Spellcheck in Emacs on OS X Mavericks with MacPorts.

Firefox

Using Firefox Dinasty I installed Firefox 144.0b5, with the help of and the links from the Mavericks Forever website.

This installs Firefox together with the tools to open websites that require a modern TLS encrypted connection.

Software from the Macintosh Garden and the Macintosh Repository

The Macintosh Garden and the Macintosh Repository are brilliant sources of software for older MacOS and Mac OS X systems.

Using these sources I installed:

Version 3.2.15 of NetNewsWire, a graphical RSS feed reader.

Version 2.0 β5 of [Notational Velocity. See my page on Notational Velocity.

Pages 5.2.2.

Karabiner

Karabiner remaps Apple keyboard keys.

I installed version 10.22 using this web-archive link: https://web.archive.org/web/20190809131630/https://pqrs.org/osx/karabiner/files/Karabiner-10.22.0.dmg

Switching the Fn-key and the left Control-key is a real game-changer, it makes using the MacBook much easier.

BTW: To use Option as Alt-key in the Apple's Terminal application (part of OS X Mavericks), check the option "Use Option as Meta key" in the preferences.

iTunes

OS X Mavericks comes with iTunes 12.6. I prefer the look and feel of the older iTunes that comes with OS X 10.3.9.

However, iTunes 12.6 is still usable. It doesn't try to convert your mp3 files to another format and it doens't require iCloud. This version of iTunes is still a very usefule tool to organise your mp3 files into an "Artist / Album / Tracks" directory structure.

VNC client

OS X Mavericks also comes with a VNC client, which is called "Screen Sharing". This works fine, and is useful to manage virtual machines running a graphical user interface.

See my page Manage remote vm-bhyve guests with OS X Mavericks VNC client.

Resources

Useful resources:

• Mavericks Forever
• The blogpost Getting Ready for the Winter. Part 1, Switching Technology
• The Macintosh Garden
• The Macintosh Repository

Kudos

Originally my idea was to have some fun with this old laptop and be done with it. Thanks you, Andreas for the inspiration go further and turn this system with OS X Mavericks into a really usable machine, and providing me useful pointers!

The Linkdump No 74 on 82MHz had a nice introduction to the blogpost Getting Ready for the Winter. Part 1, Switching Technology on the blog Minutes To Midnight by Simone Silvestroni.

In this great blogpost Simone describes several reasons to switch to OS X Mavericks and writes about favorite software recently installed on it.

Simone also points to Mavericks Forever, a wonderful website with a lot of useful tips and links for running OS X Mavericks in 2025.

I sometimes play with an old MacBook Air from 2013, running OS X Mavericks.

Inspired by Simone's post, I went to Macintosh Garden and to Macintosh Repository to find some additional software that I might like.

Notational Velocity

I stumbled upon the old version 2.0 β5 of Notational Velocity.

I have never used Notational Velocity, but I remember that several years ago a lot of people posting about being disappointed that they no longer could run it after upgrading their OS X.

In the past I have used Deft for Emacs, an Emacs mode for quickly browsing, filtering, and editing directories of plain text notes. According to the developer, Deft is inspired by Notational Velocity.

Notational Velocity can be used without a mouse, completely keyboard-driven.

Just start typing the title of a note. During typing Notational Velocity performs a incremental search for notes. It searches both in the title as well as in the body of the notes.

• To create a new note, just type the title, followed by the Enter-key.
• Otherwise, select a note from the list with search results using the up and down keys, or the mouse, and activate with the Enter-key.
• Changes are automatically saved, there is no save-button.
• To create a link to another note, just type the title within double square brackets. Follow a link with Command-Enter or by a mouse click.

Notational Velocity comes with four notes pre-installed, in one of them you'll find a list of useful key bindings.

Some useful settings:

• In Preferences: Store and read notes as plain text files (under "storage") will create text files. This way you can always use them with some other piece of software.
• In View: Switch to vertical layout. This will give a two column layout, the left column shows the search results/list of notes, and the right column the body of the active note.

Notational Velocity does what it says on the tin: it is fast and efficient to use. It is not Emacs, but just like Emacs it can be used completely keyboard driven.

The MacPorts Project provides thousands of ports that greatly simplify the task of compiling and installing open-source software on a Mac. It is easy to install MacPorts.

Once MacPorts is installed, it only takes a few commands to install and upgrade software ("ports").

MacPorts support many different versions of OS X and MacOS.

Some time ago I installed MacPorts on an old MacBook Air from 2013, running OS X 10.9.5, also known as Mavericks. This results in an old laptop running an old operating system running modern software.

GNU Emacs from MacPorts

Using MacPorts I installed GNU Emacs, it installed version 30.1, which takes quite some time because it needs toe compile software from source code.

Yesterday I upgraded MacPorts and the installed software, which again resulted in a lot of compile time, and now it runs Emacs 30.2.

Emacs on an old MacBook Air

Emacs runs of course fine on an old MacBook Air.

Unfortunately, on this MacBook the function Fn-key and the Ctrl key can not be exchanged. There is only the option to turn the Caps lock key as Control key. It is best to drop the habit of using the left Control key and only use the Caps lock key for this.

Screenshot of Emacs 30.2 in X11 on OS X Mavericks

Spellcheck on Emacs

The Flyspell package takes care of spellchecking, but it needs additional software.

sudo port install hunspell hunspell-dict-nl_NL hunspell-en_US

Next, we need to tell Emacs how to use hunspell and set a default language.

Request the location of the dictionaries with hunspell -D and add these to the Emacs init file.

(setenv "DICPATH" "/opt/local/share/hunspell/")
(setenv "LANG" "en_US.UTF8")
(setq ispell-program-name "hunspell")
      (eval-after-load "ispell"
        '(progn (defun ispell-get-coding-system () 'utf-8)))

Restart Emacs or evaluate the init file.

Check spelling with Flyspell

• M-x flyspell-buffer checks the spelling of the current buffer.
• M-x flyspell-mode enables Flyspell.
• M-$$ check spelling of word-at-point.
• C-. or ESC TAB auto-corrects the word-at-point.

For many more options, see the friendly manual.

Emacs has several fantastic functions that offer you superpowers when it comes to text editing.

Not all of these functions come with default key bindings.

To make editing prose more efficient, I added a few key bindings for transposing sentences and paragraphs.

Transpose sentences

The Emacs function transpose-sentences works great and is very useful when editing prose. By default it has no key bound to it.

First we need to make sure Emacs understands that sentences ends with a single space:

(setopt sentence-end-double-space nil)

In a brilliant post Sandra explains that in Emacs, key bindings can be different for lowercase and for uppercase chars.

We can utilize this method to construct an easy-to-remember key binding for transpose-sentences:

(global-set-key (kbd "C-S-t") #'transpose-chars)
(global-set-key (kbd "C-t") #'transpose-sentences)

The default key binding for transpose-chars is C-t. I don't use that function very often, this config moves that to C-S-t, making room for binding transpose-sentences to C-t.

If you frequently use the transpose-chars function, it is perhaps better to turn the two around, C-t to transpose chars, and C-S-t to transpose sentences.

Transpose paragraphs

Unfortunately, we can't use this method to load M-t and M-S-t with different key bindings.

Looking for a key binding that may be used within and outside org mode, I came to C-c M-t:

(global-set-key (kbd "C-c M-t") #'transpose-paragraphs)
(define-key org-mode-map (kbd "C-c M-t") #'transpose-paragraphs)

It is a little awkward to type C-c M-t, but it nevertheless works faster than M-x transpose-paragraphs.

After reading Google is killing the open web I started to style my RSS feed using XSLT.

Unfortunately, web browsers won't open a file with the extension "rss", but just offer to download it. Therefor I have to rename my RSS feed.

Old name: index.rss

New name: index.xml

The complete URL is:

http://box.matto.nl/index.xml

or

https://box.matto.nl/index.xml

For some time the RSS feed will be available on both the old and the new URL.

Please replace the old URL http://box.matto.nl/index.rss with the new URL http://box.matto.nl/index.xml in your RSS feed reader.

FreeBSD vm-bhyve is a great tool to create, install and manage virtual machines.

vm-bhyve is designed to work with ZFS, which makes it very easy to create new vm's, and make ZFS snapshots of those.

Virtual machines can either run with or without a graphical desktop. For machines running without a graphical desktop we use SSH.

The FreeBSD bhyve hypervisor provides VNC access to the virtual machines that boot with UEFI. Expose the VNC port to the network to allow for remote management.

The Screen Sharing application with a VNC connection to a Debian virtual machine running on the bhyve hypervisor

VNC access with vm-bhyve

We add some lines to the template to allow for remote VNC access:

loader="uefi"
graphics="yes"
graphics_res="1024x760"
graphics_listen="192.168.123.4"
vnc_password="secret"

• The first line sets UEFI as boot method.
• Choose a fitting resolution and use that in the "graphics_res" line,
• Set the listen address to allow remote access to the ip address of the bhyve server.
• Set a password for connecting to the virtual machine from a VNC client.

Start the virtual machine. The default VNC port is 5900. When another virtual machine is started, port 5901 will be assigned to it, and so on.

OS X Mavericks VNC client

There is no need to install additional software, OS X Mavericks has a build-in VNC client.

Open the Finder, choose "Go -> Connect to server" in the menu and enter "vnc://192.168.123.4:5900" (where "192.168.123.4" is the ip address of the bhyve server). This prompts for the password ("secret" in our example above).

The "Screen sharing" application opens with a VNC connection to the bhyve client.

The netizen club maintains a webring focused on websites that are able to be viewed in older computers.

More about the webring.

The web has become non-inclusive

Sadly, the web has become mostly inaccessible for old computers.

Websites are often only accessible using encryption. Old computers lack modern ciphers which makes it impossible to connect to those websites.

Hypertext Markup Language (HTML) is the standard markup language for documents designed to be displayed in a web browser. Unfortunately, most websites make extensive use of programming techniques like the use of JavaScript, not only making the pages slow to load, but also not suitable for older web browsers. Another disadvantage is that these techniques result in web pages not suitable for the tools used by visual impaired people.

It doesn't have to be like this

Very often the real content of a web page is dwarfed in comparison with the total web page's "weight". To publish a piece of text, there is often no real reason to use of tools like JavaScript, aside from loading web pages with surveillance advertising and other privacy invading solutions.

The netizen club webring helps to find websites that are accessible for old computers. Let's keep the web open!