Server Migration Notice (31/10/2026) — New SSH Host Keys (git.kb28.ch)
Following the server migration on 31/10/2026, the SSH host keys for git.kb28.ch have changed. If you encounter the REMOTE HOST IDENTIFICATION HAS CHANGED warning during git pull or git push, run the following command to directly replace the old key in your ~/.ssh/known_hosts file:
Official New SSH Key Fingerprints:
- ED25519:
SHA256:UrTJ+OdOKicPnL+6EgOHGlFSBm9KA+kBGMpKtoUBIOA
- ECDSA:
SHA256:qne+wTzSr8k5KIuEtXNYz9/LYed4eOMSmxzgMZ4g+KI
- RSA:
SHA256:eaVzvSUlo66pMayLMHL/cWg4zoPui3LmLOWKm0XX22g
Hi,
this PR bundles a couple of - in theory - independent fixes (See commit messages)
If you want to i can do a PR for each of the commits as well.
Best borbitter.
@@ -22,0 +18,4 @@"\\": (symbol: ">", fill: color, harpoon: true, stroke: (dash: "solid")),"\\\\": (symbol: "straight", harpoon: true, stroke: (dash: "solid")),"/": (symbol: ">", fill: color, harpoon: true, flip: true, stroke: (dash: "solid")),"//": (symbol: "straight", harpoon: true, flip: true, stroke: (dash: "solid")),Fine for now, will be fixed by updating CeTZ to 0.5.0+ (see https://github.com/cetz-package/cetz/issues/1054)
@@ -132,3 +132,2 @@let f = if i1 > i2 {-1} else {1}end-info.x -= (m.width + PAR-PAD.last() * 2) / 2pt * f(par.draw)(par, y: end-info.y)end-info.x -= (m.width / 2 / 1pt - 1.5) * fWhat was wrong about this positioning? It seems not to be aligned anymore with participants:
Hi, thanks for taking the time too look at it. Almost forgot about it.
From my Software Engineering lecture as well as this source https://www.uml-diagrams.org/sequence-diagrams-reference.html
and the official spec https://www.omg.org/spec/UML/2.5.1/PDF page 588 / Figure 17.13
I think the line should end at the left center of the top box and even be dashed apparently.
I dont know wether to change that. I do agree that this PR is in that regard somewhat a half hearted attempt at that as I didn't get it to work properly.
I will try to fix that now that I am aware of this PR once more.
Best borbitter, :)
e4f919b50dtoc92c807871c92c807871to0bc48685e10bc48685e1to45d3fd4df745d3fd4df7to3ed1966b82still confused why some file have changes that do not seem to change any graphical content.
any way I can diff them better than the gitea webui?
There should be a
diffdirectory in each test with a black and white diff image. Also, tytanic should report the number of variations (i.e. different pixels). A value < 30 is probably nothing to worry about, small text alignment issue.These tiny artifacts can occur if the Typst / Tytanic version changes. I've been working on a little tool to help me manage these version better and avoid accidentally compiling tests with the wrong Typst version
Quick reply. I would propose a tool called nix ;)
thanks for the tip with the diff dir!
I still cannot find the difference for note/styled-body/1.png
If this counts pixels correct:
magick 1.png -fill white -opaque black -format '%[fx:mean]' info:its just 1 pixel thats wrong.for group/3.png its just some minor change in outline size as far as i can tell.
View command line instructions
Checkout
From your project repository, check out a new branch and test the changes.