Website (and not only) updates

It’s that time of the year when I post website (and not only) updates, as in what changed, what improved, what is still failing, how much the Internet sucks, the society sucks, people suck, you know the drill. Personal stuff mostly.

For starters, I decided some sort of a git repository is required for my projects but I needed to provide that by maintaining the current status-quo, which means no PHP/Ruby/Python CGI support in the web server(s) and no additional server(s) ports opened except 6969, 80 and 443. And once I ran into Stagit it all made sense: static git page generator.

And now git repositories can be browsed at https://sizeof.cat/git/, git deploy is done via a custom workflow which I will detail more in a future article. Not only the repos can be browsed but they can also be cloned from the same address, because the actual git repo contents is rsynced to the web server(s), both the Clearnet and Tor ones. Feel free to inspect my git config and post-receive hooks for sensitive information, penis tasters. I mean pentesters. Or not.

And, as a bonus, repos can be cloned via Tor, too. That has to be pushing static website hosting to the extreme, right?

Due to the fact that Firefox is unusable on my Nokia Booklet 3G netbook, I decided to fork Midori (the old GTK/WebKitGTK one, not the Electron one) web browser and try to improve it a bit while getting more familiar with the Vala language. You can find my fork of Midori, called Koumori, here and you can clone the repo using git clone https://sizeof.cat/git/koumori.

Compiling it is really easy and quick (less than 5 minutes on my Nokia netbook with Intel Atom CPU and 1GB RAM, running Devuan):

$ sudo apt install cmake valac libwebkit2gtk-4.0-dev libgcr-3-dev libpeas-dev libsqlite3-dev libjson-glib-dev libarchive-dev intltool libxml2-utils
$ git clone https://sizeof.cat/git/koumori
$ mkdir koumori/_build && cd koumori/_build
$ cmake -DCMAKE_INSTALL_PREFIX=/usr ..
$ make
$ sudo make install

Speaking of the Nokia Booklet netbook, I know I failed my attempt of going back to the roots because the modern web is unusable on low-powered machines and 32bit support has been dropped from many applications (Sublime Text, Hugo, Burp Suite, Jetbrains IDEs, etc), but lately I found myself using the netbook more and more. It can’t replace my main machine for more than a few hours but it’s such a beautiful netbook.

If you have an old machine and want to have fun with it, make sure to check out the Old Computer Challenge website, and even if the challenge is done for this year nothing should stop you from joining the community.

I still believe this, you know:

Earlier this month I disabled the HTTP -> HTTPS redirect from the webserver, the reason for it is simple: I want this website to be accessible from older and underpowered machines that don’t have the computation power to encrypt SSL traffic, and besides, sizeof.cat is not a bank website. HSTS is also disabled. If you fear that you will be MITMed, exposed and flogged in public for visiting my website (keep in mind that server logs are already disabled), you might want to move to a better country. Or guillotine your current government.

I noticed there is a movement lately among the small website owners to make yourself ungoogleable. I think it was started by Joey Hess and was followed by Matthew, Andrei and Tracey. Manuel has an interesting opinion and I tend to agree: “My content is here because it’s useful for myself and hopefully for others. I don’t care about LLMs, I care about humans.” I won’t block Goooglebot for now even though the traffic from Google’s search engine is probably zero or close to zero.

That’s it. See you in 2100.