

If your filesystem is btrfs then use btdu. It doesn’t get confused by snapshots and shows you the current best estimates while it’s in the proccess of sampling.


If your filesystem is btrfs then use btdu. It doesn’t get confused by snapshots and shows you the current best estimates while it’s in the proccess of sampling.


This is the way.
Yup. Even if you add the official mozilla repos, Cannoical adds a prefix to their version so it always takes precedence over the official release. You have to pin the mozilla repo to blacklist the snapped version.
Same goes for Thunderbird.
I’m sure Snap has some security advantages for many users but they’ve made it so user-hostile for those who use native browser extensions or who want to automate deployments with just one packaging system.
Anyway, rant over - fuck Snap.
You’re going to lose Snap? That is an option, you know.


The new material has eight times better performance than pure cubic silicon carbide
Nowhere is it stated how efficient either material is, other than to say that the researchers are 5-10 years away from a material that’s 10% efficient. So they must have an efficiency of less than that I guess.
From the paper, the closest I can see is:
The applied bias photon-to-current efficiency (ABPE) of the Ni(OH)2/Co3O4/3C-SiC photoanode reached 0.47% at 0.65 V vs RHE, which is 15.6 and 1.8 times higher than that of the 3C-SiC and Co3O4/3C-SiC photoanodes, respectively (Figure 4b)
I don’t know how representative that measurement is though.


I see. mygpo is the code that runs gpodder.net. I guess it could be self-hosted, but it doesn’t look straight forward to do so. I missed it since in the docs it’s under the developer section, not the user section. gpoddersync seems much easier as long as you’re ok using Nextcloud. It would be nice if mygpo were packaged for Nix or docker. Maybe I’ll give that a go at some point.


That doesn’t clarify anything for me. Is the client application also the service, or are they (as I believe) two different things with the same name?
What I’m really getting at is that FreshRSS is self-hostable and as far as I can tell - gPodder isn’t.


I don’t think I understood what gPodder is. The website says gpodder.net is a sync service, but doesn’t seem to indicate that it can be self hosted. The list of clients has gPodder listed as a desktop PC client to gpodder.net. Does the desktop client also work as a server?
AntennaPod can sync to gpodder.net (only at that url?). When I tried it I got a load of timeouts. Instead I enabled the gpoddersync NextCloud app to my own server. That worked like a charm between AntennaPad and kasts on PC.


Thanks for posting this. I see it was on a few other communities last week but I missed it somehow. Great debugging and reminds me of the classic GTA5 online mode fix.
Does anyone know how debug symbols came to be available for the old games?


Barrier is only for inputs IIRC. To get Keyboard Mouse and Video (more usually KVM) you need some kind of remote desktop software. Rustdesk is pretty straightforward. I think Gnome handles RDP access natively now if you’re running a Gnome based Linux distro. Otherwise XRDP is a bit of a faff, but solid once it’s working.
If you have the docker-compose.yml locally, you can nix run github:aksiksi/compose2nix to translate it into a nix file for inclusion in your nixos system config. I think that could be done in the config itself with a git url but I’m not that great at nix. You will surely still need some manual config to e.g. set environment variables for paths and secrets.
How do the DNS servers resolve local hostnames then? The pihole DHCP integration adds local hostnames to DNS when they are assigned an address. If there’s two DHCP servers handing out leases, presumable only one would be accepted, how then would the DNS servers sync those names?
I think I had my secondary pihole resolve local names from the primary, and leases were copied over on a cronjob in case the secondary DHCP server had to be enabled.
Not that it particularly matters for just queries. The problem is that DHCP can only be enabled on one host. If that one fails then devices can’t get on to the network themselves. I’d like to know a good way to have a failover DHCP server - my janky cronjob isn’t great.
Where do you do DHCP? I had a primary pihole with DHCP enabled and a secondary with a cron job that enabled DHCP if the primary was down or disabled it if the primary was working. The cron job did sync DHCP leases from one to the other but it was a bit janky. I tried to update the secondary to pihole v6 and hosed it so I have no backup for now. I’d like to re-image the secondary and get a better setup - when I have time.
Edit to say I really wanted to try keepalived - that’s really cool to fail over without clients noticing.


There’s a talk here: https://www.youtube.com/watch?v=jMUMRNRkcMg but I’m not much clearer after scrubbing through some of the slides. It sounds like all GPIO pins on recent Intel chips support this Timed I/O functionality. I don’t know if they could measure the timing of transactions to a NIC - I would think this is only for specialist hardware or testing. They mention using a logic analyser to compare the clocks of two systems after synchronisation.
I think this wiki cheetsheet will explain some of the defaults. Pretty much everything is controlled with a keybind using the meta/super (windows) key. Mod + d should open the launcher, I only used dmenu but yours might be something else. The launcher will let you launch applications by name. If you just want a terminal, Mod + Enter will open one.
You will want to look at your config. It should live in ~/.config/sway/config. If it’s not there then mkdir ~/.config/sway/ && cp /etc/sway/config ~/.config/sway/. That should list the keybinds you have set. You can look up the options in the man page for sway.
Once you can do some basic window control, you might want to customise the status bar. The config should tell you what bar is being used, but there are a huge array of statusbars to choose from - I used i3status-rust but try searching for i3/sway statusbars to see what’s out there.
Ah, I saw another comment about this. The free plan is 300,000 queries a month. That’d last me almost a week before it stops working.
That says it will only function for 300,000 queries per month. Based on my last 24 hours from pi-hole, that wouldn’t even last a week. Are you using a paid plan?


Yeah, I really dislike snap and have puppet clean it out and add in the real mozilla repo for me. If I wanted sandboxed apps I’d probably look at flatpak but I think there’s still work to be done there also.
They haven’t modified
apt; they abuse an extra version number that supercedes the major version number of a package. I think it’s meant to be used for new packages that reuse the name of an abandoned project. Canonical publish packages for software like Firefox that depend on snapd and just runsnap install firefoxinstead of actually installing anything. Since they bumped that extra version number, their packages always have a higher precedence than even the officially packaged debs from Mozilla.