The goal here is to put together a portable rust/gtk4 development environment on windows. As of this writing I’m using windows11.

Install Choco

These days I’m using the choco package manager on windows. Installation instructions here

Install Rust and msys2 (using Choco)

$ choco install rustup.install msys2
Chocolatey v2.7.4
Installing the following packages:
rustup.install;msys2
By installing, you accept licenses for the packages.

Update rust with

rustup update

Add the GNU toolchain

The Doko page suggests doing this even before installing the gnu tools. I had my doubts but it worked.

rustup target add x86_64-pc-windows-gnu

Run …

rustup show

… to show the current target:

stable-x86_64-pc-windows-gnu
stable-x86_64-pc-windows-msvc (default)

Run …

rustup default stable-x86_64-pc-windows-gnu

… to switch targets and run …

rustup show

… again to see that the current target has changed.

Using a MSYS Shell, run the following commands;

pacman -S mingw-w64-x86_64-gtk4
pacman -S mingw-w64-x86_64-toolchain

The second (toolchain) pacman command gives some options. The necessary ones for this task are: “binutils”, “gcc”, and “pkgconf”. Or take the default to install them all which takes longer but seems to work.

Add mingw64/bin to your PATH

The following is from my shell start script. I’m running bash in cygwin but something similar should work in the windows cmd.

export PATH=\
/cygdrive/c/tools/msys64/mingw64/bin:\
$PATH

It seems that this is necessary to give access to an executable, dlltool.exe, as well as some other things that are found based on the location of dlltool.exe. If you see an error message related to dlltool not being found, check your PATH.

Compile and run gtk4 programs.

Examples are available as part of the gtk4-rs repository. They work both on Windows and on Linux without modification.

diagnosis and summary thanks to Claude

Symptom

On this Dell 14 Plus 2-in-1 (Intel Lunar Lake, Arc 130V/140V graphics), the screen occasionally goes completely black and the keyboard stops responding — except F5 (keyboard backlight toggle) still works. Happens after long periods of inactivity with the laptop lid open (not closed), on battery power. Connecting/disconnecting the charger while stuck made no difference.

Diagnosis

Checked journalctl/dmesg across ~10 recent boots. No kernel panics, GPU hangs/resets, or unclean shutdowns were logged — but a clear pattern emerged:

• Aug 23, 19:17:06 — system entered suspend (s2idle) after the 15-minute battery idle timeout
• ~20:34–20:35 (≈78 min later) — kernel and networking resumed successfully (confirmed via Tailscale reconnect logs scrambling back online at that exact time)
• Display and input never came back — nothing further logged until a hard power-cycle forced a new boot

Root cause: the kernel resumes from suspend correctly, but the xe gra driver / Wayland compositor fails to reinitialize the display pipeline afterward — a known class of bug on this very new Lunar Lake + xe driver combination. Backlight toggling works because backlight toggling is handled by the EC/ACPI layer directly, not through the wedged compositor. BIOS was already current (1.11.0), so this is a driver-side issue, not a firmware gap.

Fix Applied

Disabled auto-suspend on battery so the machine can’t hit the broken resume. It will still blank/lock the screen on idle, just won’t suspend:

gsettings set org.gnome.settings-daemon.plugins.power sleep-inactive-battery-type ‘nothing’

Verified current settings:

Safe Recovery (if it happens again before/without this fix)

Instead of holding the power button (risks filesystem corruption from unflushed writes), use the Magic SysRq key — handled directly by the kernel’s keyboar interrupt handler, bypassing the frozen compositor entirely.

This system’s SysRq is enabled at level 176 (Ubuntu default), which permits only:

• S — sync disks (flush dirty pages)
• U — remount filesystems read-only
• B — reboot

(R/E/I — raw keyboard toggle and process terminate/kill — are disabled by default on Debian/Ubuntu and won’t do anything here.)

To use: hold Alt+SysRq (often shares the PrtScn key, sometimes need tap S, release, tap U, release, tap B — with a brief pause between each.

The two posts accomplish the same thing but the steps in the earlier post, worked with the repository containing the jekyll source. The steps in this post work by creating an additional local repository in the _site directory that contains the content to be published.

If you publish a Jekyll based website on Github they will do the Jekyll compile for you. This is easy and quick to set up but here are some reasons you may want to build on your local machine and push the results up to Github:

• Github only supports a limited number of Jekyll plugins. If you use any plugins outside of that set then you must build locally and push the result.
• Other web hosting sites will not support automated Jekyll builds in the same way. If you want to compare sites you may need to do a local build anyway.
• Github does not provide automated builds for static site generators other than Jekyll. Knowing how to push a local build lends itself to working with those other static site generators.

Here’s a script that provides some help in pushing a local build to the gh-pages branch for a repository. It’s not very polished - users will need to edit it to specify the correct baseurl for the jekyll build.

#! /bin/bash

# -- Build with baseurl, '/ntest'
# bundle exec jekyll build --baseurl '/ntest' 

# -- Build with no baseurl
bundle exec jekyll build 

# -- Switch to the _site directory
cd _site/

# -- Set up a git repository for this directory ( if not already done )
git init 
git branch -m gh-pages
git remote add ghp git@github.com:dc25/davecompton.net # of course, use your own remote repo 

# -- add, commit, and push _site contents to gh-pages branch.

git add .
git commit -m 'adding site to be pushed to gh-pages'
git push -f ghp gh-pages

After pushing to gh-pages, github will unset any custom domain name you might have specified. Just re-enter and save it if needed.

This post and the technique described here are thanks to claude.ai .

Prerequisites

sudo apt install open-vm-tools

Steps

1. Create the unit file

sudo tee /etc/systemd/system/mnt-hgfs.mount << 'EOF'
[Unit]
Description=VMware Shared Folders
DefaultDependencies=no
Requires=open-vm-tools.service
After=open-vm-tools.service

[Mount]
What=.host:/
Where=/mnt/hgfs
Type=fuse.vmhgfs-fuse
Options=defaults,allow_other,auto_unmount,entry_timeout=0,attr_timeout=0

[Install]
WantedBy=multi-user.target
EOF

2. Enable and start it

sudo systemctl daemon-reload
sudo systemctl enable --now mnt-hgfs.mount

3. Verify

systemctl status mnt-hgfs.mount
ls /mnt/hgfs/

Notes

• DefaultDependencies=no is required. Without it, systemd creates an ordering cycle: .mount units are automatically added to local-fs.target, but open-vm-tools starts after local-fs.target, so the mount can never satisfy its own dependency. DefaultDependencies=no removes the unit from local-fs.target ordering, breaking the cycle.
• An /etc/fstab entry is not a viable alternative — fstab-generated units have DefaultDependencies=yes and there is no fstab option to override it.
• The mount point /mnt/hgfs does not need to be created manually; the fuse driver creates it on first mount.
• Shared folders appear as subdirectories under /mnt/hgfs/ (e.g. /mnt/hgfs/shared).

Mafl LXC

Installation went smoothly. Access was in browser via port 3000. This results in a page that looks like this:

As far as I can tell, each of these icons is a link that by default is not enabled. I looked briefly at the documentation to find out what it takes to get them working but got impatient and moved on.

Heimdall Dashboard LXC

Again, install went smoothly. Access through port 7990, Used the builtin gui to add a couple of “apps” resulting in this:

These are clickable and I configured them to link to the plex and pi-hole web interfaces.

I’m starting to get the impression that a “dashboard” is just a fancy collection of bookmarks.

Homepage LXC

Again, install went without a hitch. Access using browser at port# 3000. Configure using scripts at /opt/homepage/config . I added plex and LMS to get a setup like this:

Thoughts

These seem pretty easy to install and configure but I don’t think they buy you all that much. I’ll probably stick with plain old bookmarks.

$ choco install openssh
Chocolatey v2.2.2
3 validations performed. 2 success(es), 1 warning(s), and 0 error(s).

Validation Warnings:
 - System Cache directory is not locked down to administrators.
   Remove the directory 'C:\ProgramData\ChocolateyHttpCache' to have
   Chocolatey CLI create it with the proper permissions.

Installing the following packages:
openssh
By installing, you accept licenses for the packages.
Progress: Downloading openssh 8.0.0.1... 100%

openssh v8.0.0.1 [Approved]
openssh package files install completed. Performing other installation steps.
The package openssh wants to run 'chocolateyinstall.ps1'.
Note: If you don't run this script, the installation will fail.
Note: To confirm automatically next time, use '-y' or consider:
choco feature enable -n allowGlobalConfirmation
Do you want to run the script?([Y]es/[A]ll - yes to all/[N]o/[P]rint): Y

Running on: Windows 10 Pro, (Professional)
Windows Version: 10.0.19045

************************************************************************************
************************************************************************************
This package is a Universal Installer and can ALSO install Win32-OpenSSH on
Nano, Server Core, Docker Containers and more WITHOUT using Chocolatey.

.... (more output)

Now, per OpenSSH instructions , open port 22 , install openssh as a service, and start the service

dc202@DESKTOP-MAR8MAD /cygdrive/c
$ cd 'c:/Program Files/OpenSSH-Win64'

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files/OpenSSH-Win64
$ netsh advfirewall firewall add rule name=sshd dir=in action=allow protocol=TCP localport=22
Ok.


dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files/OpenSSH-Win64
$ powershell.exe -ExecutionPolicy Bypass -File install-sshd.ps1
[SC] SetServiceObjectSecurity SUCCESS
[SC] ChangeServiceConfig2 SUCCESS
[SC] ChangeServiceConfig2 SUCCESS
sshd and ssh-agent services successfully installed

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files/OpenSSH-Win64
$ net start sshd
The OpenSSH SSH Server service is starting.
The OpenSSH SSH Server service was started successfully.


dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files/OpenSSH-Win64
$

That seems to be almost enough but when I try to log in remotely, ssh rejects my password:

dave@vbox:/tmp$ ssh -l dc202 192.168.8.146
dc202@192.168.8.146's password:
Permission denied, please try again.
dc202@192.168.8.146's password:

At this point I searched online a little for a solution but eventually gave up (for now).

I restored the machine to its previous state. Thanks to this superuser.com answer from “Vomit IT - Chunky Mess Style” for telling how to undo the firewall changes.

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files
$ net stop sshd

The OpenSSH SSH Server service was stopped successfully.


dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files
$ netsh advfirewall firewall show rule status=enabled name=all | grep sshd
Rule Name:                            sshd

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files
$ netsh advfirewall firewall delete rule name="sshd"

Deleted 1 rule(s).
Ok.


dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files
$ cd 'c:/Program Files/OpenSSH-Win64'/

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files/OpenSSH-Win64
$ powershell.exe -ExecutionPolicy Bypass -File uninstall-sshd.ps1
sshd successfully uninstalled
ssh-agent successfully uninstalled

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files/OpenSSH-Win64
$ cd ..

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files
$ choco uninstall openssh
Chocolatey v2.2.2
3 validations performed. 2 success(es), 1 warning(s), and 0 error(s).

dc202@DESKTOP-MAR8MAD /cygdrive/c/Program Files
$ rm -rf OpenSSH-Win64/

If anyone reads this and happens to know what the missing link is to get logins working, please let me know.

• For reddit.com/r/rust RSS feed, use : www.reddit.com/r/rust/new/.rss

• For stack overflow questions tagged rust use: https://stackoverflow.com/feeds/tag/rust?sort=newest

• For google alerts, create the alert via alerts.google.com, edit the alert to “Deliver to” RSS feed, then follow the RSS icon for the new alert.

• For mastodon, use, for example, https://mathstodon.xyz/@johncarlosbaez.rss

• For twitter, use a nitter instance. For example: https://nitter.privacydev.net/nytimes/rss . Twitter seems to be a moving target.

  • As of 8/21/2023, the twitter/nitter technique here worked but it may not last.
  • As of 09/01/2023 it’s broken.
  • If anyone reading this knows a way to view a twitter feed as rss, please let me know.

Dropbox:

How to get dropbox image url.

Flickr:

How to get flickr image url.

github:

I got the image link by :

• Uploading the image to a github repository.
• Clicking on the image in the repository
• Right clicking on the download link and selecting “Copy link address”

imgur.com:

Just click on the uploaded image in imgur.com and use the “Direct Link”

imgbb.com:

use the “Direct links” option in the “Embed codes” menu for the uploated image in imgbb.com

Below is the markup for this page, included to show how to embed the images.

Later on, working in Ubuntu now, I searched the web and found these instructions that say to

• change /etc/default/grub to specify a default
• run update-grub.

I did this but nothing changed.

The reason that nothing changed is that Mint had taken ownership of the boot process. After booting into Mint and making the same changes, the default boot option was updated as expected.

Looking back, this should have been obvious but it cost me a few minutes.

At some point, I expect that an Ubuntu update will again rewrite the grub configuration and I’ll have to go through the process again on the Ubuntu side.