The past year I’ve been playing pretty active on federated social networks. This led me to try and write a federated social network group system. Although this group system is far from being complete since Fedigroup is currently in pre-alpha level. I think it is worth sharing what I have learned, and also explain a bit why, if you met me during the past three years. I’ve been going about federated social networks.

You are also welcome to join a server I maintain here, Hayu.sh is the second largest Hebrew-speaking instance.

What is a federated social network?

A federation in servers means everyone can run their own server which is autonomous. And it can communicate with other servers too. Creating a large network. The first protocol on the internet is actually federated too – email.

There is no one central email server, or single email software. In social networks this means that you join a server, and follow and interact with everyone else. Even though your server could be, like in my case, hosted in your own home, with you owning all the data. You can also join a any other server, and switch between them.
For Hebrew speakers, I gave a small TED lecture about it, which is hosted on a federated video server. You can watch it here (Hebrew).

The Fediverse is the servers using the a W3C Standard that does just that

What made the Fediverse possible is the standard that was passed to describe it. The protocol is called ActivityPub 2.0. And it lets social networks share status messages (notes), shares (boosts) and favorites across different servers. It was created by a group of people who cared, not large companies. Despite some of them being members of the W3C. Its also worth mentioning that one of the major adopters of the ActivityPub 2.0 protocol was Mastodon, a type of social network server, which resembles twitter. And that really help gain critical mass.

Why didn’t I hear about this?

I was asked this by other developers and power users. It makes people dismiss the Fediverse as a passing or niche thing.

Unlike Facebook, Twitter even if you’re been avoiding social networks, their marketing will reach you, your news channel, municipality and country are likely using them. Their existence depends on it that you use and browse on their site. However, in the Fediverse does not have a business model that demands it. Or a marketing team. Its just people. You are welcome to join, but no ones income depends on it.

The Fediverse is growing though, by word-of-mouth, or text-in-blog. Recently even celebrities and others. The latest are the BBC, PNAS, George Takei, Neil Gaiman, Stephen Fry and the European Commission.

Not to be confused with crypto – Which lacks moderation

I think one of the strength of the Fediverse is that it’s actually not going with the blockchain trend now, which is where a lot of effort is going to distribute things.

The Fediverse does not run using the now trendy crypto stuff. The advantage of it is that it costs far less to run. A small server for a few tens of people + a domain and you are good to go.

Another thing is that having an admin means you have a moderator, which is becoming critical for social networks. Large companies today are unable to moderate the volumes of messages on their social networks, resulting in many bad decisions, people getting blocked by algorithms and others who should roam freely. Many social networks today still don’t have moderation-by-design in mind. Even less in a trust-less network that uses crypto.

What is Fedigroup

Fedigroup is a server type that allows users to create a group and share it across the Fediverse. When a user creates a group, they can specify the topics they want to discuss, and the group, when mentioned, will automatically boost their posts related to those topics. This helps increase visibility for posts within the group and attract more members. Additionally, Fedigroup stores the contents of the group for future reference, making it easier for members to find and access previous discussions. Currently, there is also a web interface available that allows users to search through the contents of groups they belong to (the search button though is not implemented yet). Overall, Fedigroup provides a convenient way for users to create and participate in specialized communities within the Fediverse with group memory.

Features

• Federated – you can mention a group and it will boost the message if you are following it
• Has a responsive UI written in NiceGUI
• You can log in using your Mastodon/Pleroma or any MastodonAPI compatible account
• You can use your federated account to create group and then you have access to moderate it
• Posts in the group are archived in a database that can be searched (search not implemented)
• See all the posts in the groups you follow

Unimplemented (roadmap in theory)

• Use your MastodonAPI-compatible account to comment and preform tasks on the WebUI
• Search history of the group

I stopped developing it because it was far too much work only for me. It might pick up if someone joins me, but alone I will not manage to make it as good as others out there. It has potential as being a place to search for history on the Fediverse which is a little challenging today.

Also I think the recent migration of Reddit users show that the Fediverse is handling groups pretty well.

Still, Fedigroup was a good learning experience on how the technology works, and I think its worth sharing if you want to build a Federated instance that uses python, FastAPI and PostgreSQL.

If anyone wants to pick up where I left they are welcome. The source code is available on GitHub.

There is a running instance at: https://fedigroup.party
You welcome to try it out and as always – code contributions are welcome

Hey all,
I am releasing today something I’ve been working on the past year. It stared with a friend that had a requirement to do off-site backups of his work for his insurance, but BackupFriend has became a full blown backup solution.
The idea is you get a RaspberryPi and plug a harddrive to it, put it at your friends, family or just in your house and backup up to it using a desktop client that should have the feel like Dropbox/Google Drive/etc. You also get history! The RaspberryPi can also be changed to a remote server (instructions here). This guide focuses on self-hosted instructions, because I think its strange all backup solutions out there require you to pay rent for your own data.
Using a 4TB drive for under $100 and a Pi, you can get much more storage which is not ephemeral.
The desktop client has been tested on both Linux, MacOS and Windows.

Also, many thanks to Pull Request Community that featured BackupFriend as a project to work on, that helped get a few PRs on-board.

What you will need

1. A Raspberry Pi, preferably 4B because of the USB3
   • Note: Raspsberrypi can be replaced with a server (instructions here).
2. A home network with a computer running Linux, MacOS or Windows, which you will be backing up from
3. An external hard drive that can connect to the Pi (I use a 4TB WD My Passport drive, you can also just use a thumb drive for testing).

How to install on a Pi

1. Download and the BackupFriend image (source code so you can see what’s under the hood)
2. Unzip the image and install it to an SD card like any other Raspberry Pi image
3. Configure your WiFi by editing backupfriendpi-wpa-supplicant.txt at the root of the flashed card when using it like a flash drive (You can also just plug the device to an Ethernet port on your router).
4. Attach the external drive and power up.
5. Hostname is backupfriend (not raspberrypi as usual), username: pi and inital password is: raspberry
6. After a few minutes you should be able to access http://backupfriend/ or http://backupfriend.local.
7. Default password for the Web UI is admin password admin123.

You should see a login screen like this:

At this point you have the backup server running. Excellent!

The backup server is now sitting on your local network. If you want to have it backing up on a remote location, you will need to open the router port to it, and set the domain name on the nginx-proxy container there. More info on that here.

How to set up the Desktop Client

Now that we have a server on the Pi, let’s backup to it form the local host.

Linux

sudo python3 -m pip install backupfriend

Windows

1. Download .zip for windows run the executable backupfriend-client.exe.
2. If you want to make it start on boot follow this guide.

Mac

1. Download the .zip for mac and open it to extract the .app
2. Copy it to our applications folder.
3. If you want to make it start on boot follow this guide (guide would not work if the .app is not in the applications folder).

First Time Run

On the first run you will be asked to follow a wizard and create your connection keys. Those let you backup sequentially to the server. Just follow the guide and copy the public key:

Server setup

Now that we have a key, let’missing apostrophes place it in the server.
Log in to your backup server using the default

user: admin
password: admin123

You should get this view as shown below

Go to admin on the top right and select “Profile”

We need to set up the sync folder and SSH key.

In the SSH key page press “add new SSH key” and paste the public SSH key you got from the BackupFriend Client:

Now there is one more step to do before we can start backing up – setting the root folder for the user.
Click on “Admin area” and then on the “Users: 1” button as shown below:

Select the only admin user in the list and click on the “edit” button. You should see the following edit page.

Here you need to add the “Root directory” to be “/backup”. This means that anything in the “/backup” folder on the Pi would be visible in the web interface.

Once the changes are saved you will see the amount of free space on the main window:

Now we can finally sync. The first install comes with an example sync. It just syncs your temp folder. Click the “edit” and then you can confirm that it all works using the “test” button. You should get “connection succeeded”.
Note: Sometimes you might see first “warning: server added to authorized servers”. Just press “Test” again, its the SSH system setting that server as an authorized one.

When the job is running it will turn blue, when it’s finished it will turn green:

Once it’s done, or in the middle, you can access the backup on the web interface.

To do so you might need on the first time to refresh the repositories to it will be shown. To do that go to the “profile” and press “refresh repositories”:

Finally, you can see now the repositories and the files with their versions:

That’s it, you should have now a backup server on a Pi that you can also move elsewhere.

Development – Source Code Repos

There are quite a few parts for the project, here are the repos for them:

• BackupFriend Client – wxpython mostly)
• BackupFriend-docker – server-side Docker builder)
• BackupfriendPi – The RaspsberryPi image builder
• Rdiffweb fork – The web interface fork that runs on BackupFriend

This is just version 0.1.0 and there is a lot to improve. Since I am only one person I would greatly appreciate PRs and help getting this to be a full blown solution.

Hey all,

A short post to announce a small piece of software I released today on github for the ham radio people.
I give you SvxlinkWrapper, this wrapper sits on top of Svxlink‘s stdout and stdin and processes it with python. Making it easy to add more advance features to Svxlinik. Since Svxlink is written in C++ and takes a while to compile, this wrapper really speeds up development. Furthermore, it uses straight.plugin making its simple to write more modules by extending SvxlinkwrapperModule and placing it in the modules folder.

Currently the to modules that SvxlinkWrapper has are for auto-connect to Echolink nodes on startup and another for a QSO logger for Echolink connections. There is also a module that should let you send commands over the Echolink chat, if you extend it. This lets me open and close repeater connections directly from the Echolink chat box.

• Get SvxLinkWrapper
• Example of the logger on the Jerusalem Repeater R1

Thats it for today, now that I have more free time, I hope to publish more things, 73

TEDing from the car

Hey all,
After a few weeks of bouncing emails with Thomas Perl I am happy to announce that we have a working plugin for his wonderful podcast application called gPodder. With this plugin installed, any mp4 video podcast that is downloaded with gPodder is automatically converted to a format playable by rockbox MPEG player plugin. Effectively meaning seamless sync of video podcasts to any rockbox enabled device. I am specifically using the Sansa fuze player, but just changing the screen resolution at the top of the plugin should make it work with any other rockbox device, and with the right screen size. The final product for me of all this technical stuff is that I can listen to TED lectures in the car. Or using the verb was coined by my friend Uzair “TEDding” from my car.

How to Install

• First you should have gPodder installed (available also as “apt-get install gpodder” or any other distribution install equivalent).
• Second you will need these packages installed too:

   apt-get install python-kaa-metadata  ffmpeg python-dbus

• Now all that is left is to copy this script to

  mkdir -p ~/.config/gpodder/hooks/
  cp rockbox_mp4_convert.py ~/.config/gpodder/hooks/

• If your are using a different player from the Sansa Fuse, modify the first lines to your screen resolution:

  DEFAULT_DEVICE_WIDTH = 224.0
  DEFAULT_DEVICE_HEIGHT = 176.0

  make sure to include the .0 at the end.

• Thats it! Now each time video is download, for example from the TEDTalks Video Podcast it would be converted on arrival, and relinked as the file to be synced.

As usual, I would love to hear feedback from users and suggestions.

TEDing from the car, thanks to a gPodder hook and Rockbox

gPoder downloading and converting

SMS Twitter Gateway logo

Hey all,

I am releasing here set of instructions and source that will enable anyone to set up a phone number, that if you SMS it, will post your massage on to your own twitter account. This service is called “Twiter SMS Gate”. The service handles multiple users, and should be easily modified to support other services like statusnet (and with a bit more code maybe even other social networks). The SMS hardware required is any cheap cellphone that can work with gammu-smsd (mine is a fake Nokia from India). Twiter SMS Gate also lets users easily register to it with their own cellphone.

I hope that people that like this idea would start hosting more SMS gates, as we create a wide network of those around the world. SMS Twitter Gate-IL , The Israeli Gate, has been running for a few weeks now and even got covered in the local media here.

The source is written in Python, which includes the phone hardware hooks, webserver, twitter client and database access.

Here is the cellphone used to run my server, just to show you how simple it is to do this, no super-expensive hardware required! (server is also my personal Linux-running PC:

Fake Nokia Cellphone acting as a dedicated SMS server

• Source is available here on GitHub, along with basic instructions on how to set it up.
• Working example of an SMS Twitter Gate in Israel
• List of working gates (one at the time of writing this)

If you have any kind of difficulty setting this up, please give me feedback, so we can make it as easy as possible.

3arabi Logo

Hi  all,

As some might have noticed, Arabic speakers on the net use a form of writing called ‘Arabic chat‘ or 3arabi, which involves using Latin characters and Hindu-Arabic numerals to write words in Arabic. I wrote a small service called 3arabi that lets you translate this Arabic chat directly to English.

Apparently there are tools to do transliteration (converting Latin letters back to Arabic), and also translation. But nothing that does them both. That is why I wrote a small script that uses Google’s transliteration service and ‘Google Translate’. This does the job, but is not perfect, however, it does actually help me understand some messages.

The source is also available in GitHub (its in python). If anyone contributes better code I’ll merge it back to the service.

• 3arabi translation website
• Here is an example of linking to a translation

Thanks to Ira Abramov for hosting the service!

Thanks to Rajesh who wrote the transliteration API though after using it I moved to the javascript google tool with his help.

Thanks Muhammad Khatib who wrote Google Translate python API and for releasing it.

Enjoy,

Guy

Update: Google seem to be blocking the translation service, thinking its spam, if anyone knows how to contact them and asking them to lift that ban would be appreciated.

CHDK used in university for capturing blackboards

Hi all,
During last semester I developed a set of scripts that lets me photograph pictures of the blackboards during my university courses, categorize them while I am taking them and automatically generate PDF document for each course week. Because physics equations are not the simplest things to input in to a laptop in realtime, I have developed the following method. I am writing the post because I have a feeling this tool could help people in many ways, since it did change how I study in lectures.

You can see my blackboard photographs here (and here is a simple pdf example, if you get lost in the Hebrew).

How it works in a nutshell

The work flow - photograph, store, generate pdfs

During the lesson I take the photographs of the blackboard with a script I wrote for CHDK. In the script you set the course and week of the semester and this is stored per-picture on the SD card.

When I get home, I can quickly clear the card using the build.py script, moving the pictures to my computer, I might be taking dozens of photos a day, so its useful to do this a few times a week. The images are automatically stored in corresponding folders (e.g. /camera/coursename/pics/week1/ ).

Once the images are in the directory structure, all that is left is to generate PDFs from each ‘week folder’. There are quite a few things taken in to account here, due to the fact that the images are resized for fast PDF rendering, to setting the temporary folder of Imagemagick (it needs a few gigs to generate the PDFs!). The final outcome is a PDF file per-course per-week. From here I will go in to the technical configuration.

What you will need

• Canon powershot camera running the CHDK software (Cannon Hack Development Kit). I use a Cannon powershot A590
• Python + bash (+ Linux)
• Imagemagick installed (the convert command)
• Free space is important (both for PDF generation and storage of your photos)

Setting up CHDK script

I will not go in to the details of installing the CHDK, I wrote about this already, and its a post I recommend you read!

Once you have CHDK running on your camera you will need to use this script for shooting pictures (copy it to /CHDK/scripts and load it).

The script lets you pick the course and week, you can edit the script to have a little abbreviations for courses. For example (line 3 in blackboard.lua):
@param c Course (0=inf,1=QM,2=ds,3=fn,4=op)

It will give you a message each time you photograph with it.
The script saves files with the images filenames to the LOGS folder on the card, containing the course ID and week number.

Screenshot of the CHDK script menu

Importing you photos from the camera automatically

Using the build.py will move all the photographs from the camera to the disk, you will need to set a few things in its header, here is how the build.py file looks:

### configuration ###</p>
<p>#path to which the camera is mounted<br />
cameraMount=&quot;/media/CANON/&quot;</p>
<p>#path were the photos should be moved to<br />
DESTDIR = &quot;/media/Elements1/University/camera/&quot;</p>
<p>#path where the photos are in the camera (you might need to change this if you took more than 1000 photos)<br />
PICDIR = cameraMount + &quot;DCIM/100CANON/&quot;</p>
<p>#the number, and hte course folder to move it to<br />
course= []<br />
course.insert(0, &quot;infi&quot;)<br />
course.insert(1,&quot;quantum1&quot;)<br />
course.insert(2,&quot;dast&quot;)<br />
course.insert(3, &quot;lab2&quot;)<br />
course.insert(4, &quot;oop&quot;)

Its pretty straight forward. The last list sets which ID number corresponds to which course.

Generating PDFs

The script to generate the PDF is called genpdf.sh, It basically scans the folder and generates the pdf files using imagemagick, it also creates a small file with the picture list, so it only generates when there is a change. Generally useful script! I recommend you set the temp folder that it uses for the generation to be somewhere on your disk (and not just /tmp), because it can take as much as 10GB sometimes to generate!
Here is the settings header:

#this temp dir is where the PDF are generated, it can reach gigabytes! so its recommended to set it to a place with space<br />
export TMPDIR=/tmp</p>
<p>#path where the images are saved<br />
PICS_DIR= /media/Elements1/University/camera

Camera setting tips – How not to blind your lecturer

If you are using this tool, here are just a few minor tips for getting a good result:

• First use the manual setting in the camera (the M option). If the course is in a closed lecture hall you can tweak the exposure time to get better resultes. As you can see I also set the picture to grayscale.
• Make sure to set the flash off! Also you can set the AF-Assist beam off (its the red light on the camera). The manual setting remembers these options, so this is useful too.
• I personally try not to photograph lecturers,  I suggest you do the same.

What courses are good for this method? – Tips for Lecturers

I found that some courses are better than others to photograph, usually if you have a tidy lecturer who doesn’t erase in the middle of calculations, then you tend to get a good result. Also there are some lectures that think that the whole lecture can be done with equations only, no text. Those lectures tend to be much harder to follow, I would really suggest that they add notes, it tends to help everyone. Even a tiny title over a calculation can really change things, students don’t always follow.

That’s it! This technique really changed how I study in university, I am sure there are ways to improve the process and the code or use this for another things. I would love to hear feedback and would be glad to assist anyone attempting this.

• Here is a package with all the scripts you need

Hi all,

During my small semester break I managed to connect my Bumble-b chip to an analog-digital converter (ADC).

With this I can read voltage changes and send the information to the computer to plot live.

Here is a demonstration video:

How its all connected

The Bumble-b is connected to the MCP3204 ADC, here is the schmatic from dflecher’s site:

MCP3204 wiring schematic

MCP3204 wiring schematic

Software on the Bumble-B

The pulling is done with a library written by dflecher and help from LUFA (The USB-AVR library). My source is available here I recommend you look at my original Bumble-B post on how to compile and develop.

Software on the computer

To plot I used a pyOpenGL, which is a python binding to draw graphics. I based it off one of the demo examples for pyOpenGL. I did not spend much time on it, and there is a lot to improve. Like adding a grid, more options and speeding up the incoming data. Still, this tiny script is a proof of concept.

Small basic oscilloscope showing a remote control signal

Small basic oscilloscope showing a remote control signal

I would be glad if anyone tells me how to speed up the Bumble-b serial communication, and draw faster on PyOpenGL, since this is the current bottleneck for a faster oscilloscope.

Here are all the sources I wrote for this post:

• Source for the bumble-b driver
• Source for the pyOpenGL live plotting scope

Note: This project is also compatible with the upcoming Bumble-b2

That is all for now, I hope to find time to post during this long semester

A while ago, when I coded message alert plugins for pidgin and X-chat so each time I get a new message my bumble-b LED display would flash, so I thought I might release the code here for people to use.

Both plugins will execute a command of your choice when receiving a new message.

Thanks to Simo Mattila who wrote the base code for the pidgin plugin

• Source code for the pidgin plugin
  • You will have to follow the README file and compile the plugin. Then place the created .so file in ~/.purple/plugins/ .
  • In the plugin menu there would be a configuration window to set the command.

• Source code for the X-chat plugin
  • You will have to set the command inside the python file. Then copy it to ~/.xchat2/
  • The X-chat plugin also is able to return the name of the user that sent the message.

Here is a video of the X-chat plugin in action:

I hope to find more time to posts the things I am doing, but as usual, university takes up all my time. Stay tuned

The effect is that I can publish all of my video podcast to youtube, without the long and painful chopping and uploading of files via a browser.

I based the uploading on a perl script from catonmat. I wrote my extension in python.

Here is a link to the sourcecode (under GPL license)

I would love to receive feedback, and improvments.