The Arch wiki may have some ideas for you - tl;dr is that GDM uses a global dconf db over in /etc/ and this might be the root of your problem (these configs might not get cleaned up with a --purge?) I’m a LightDM user so best I can do to help: https://wiki.archlinux.org/title/GDM#dconf_configuration

Quick update for anyone still reading this thread:

@[email protected] As with any other app, we flagged Fennec and Mull with KnownVuln until the app is updated. Contributors fixed the issues that delayed versions 130 and later. Stand by for the build.

https://floss.social/@fdroidorg/113384089915217604

A bit of backstory on how we got here - in June 2024 Mozilla chose to (a) integrate the source tree of Firefox Mobile into their huge monorepo (“gecko-dev”), and (b) move the source off of Github onto their own git servers (“Mozilla Central”). You can read about it in the now-archived old repo:

• https://github.com/mozilla-mobile/firefox-android
• https://github.com/mozilla-mobile/firefox-android/wiki#upcoming-migration-to-mozilla-central

This was then compounded by a core Android build kit (“NDK”) choosing to remove parts of the toolchain which is/was used to build Firefox releases (ergo, forcing another change to build process):

• https://github.com/android/ndk/issues/2010

Together these have caused a bit of a kerfuffle in getting new releases compiled and released via the official F-Droid methodology. See the other comment about the Mull version in their private repo, they’re having to use a Mozilla pre-built clang (a compiler toolchain) now to make it work for the time being.

The link(s) to add their F-Droid repo if not running DivestOS: https://divestos.org/pages/our_apps.html#repos

The other data shows that posts and comments are going up linearly (a little suspicious but OK), but I wonder how the modlog affects the data (meaning how is it captured and when). I made one comment to a honest post yesterday (hosted on a remote instance), which then the post was deleted by admins like so:

Removed Post Any app for call recording ? reason: Rule 2: Please use [email protected] for support questions.

So my comment shows in my history but cannot actually be accessed; was this comment counted? was that post counted? Was I counted as an active user yesterday if that was the only activity I did all day? Was the one person who upvoted my comment before the thread was deleted counted?

Lies, damn lies and statistics. :)

Most of them (besides weechat-android and quasseldroid which use bouncers/relays) seem to have fallen out of maintenance; Goguma appears to be currently maintained and updated as a pure standalone client and would be what I’d recommend trying first.

I have been using Linux on laptops as main/only compute since around 1997 (started with an Inspiron 4000, PII-400 IIRC), Dell is generally extremely boring and very Linux/BSD compatible. I have been buying gently used Precision models (typically using local marketplace, Craigslist in USA) as they tend to have better build quality and non-janky custom parts (think “winmodem”). They last forever, pretty much every Linux/BSD distro works. The most important thing is to stay away from Broadcom chips and look for Intel eth/wifi. Stay away from Inspiron to avoid hardware problems, in modern times those are the bottom of the barrel janky hardware.

The Dell Latitude line used by businesses are even more boring than Precisions and really always have been - their BIOS has a somewhat unique charging profile “always plugged in” to extend battery life - I use two ancient E6330 models tuned to super low power modes as mini-servers (think anything you’d use a raspberry Pi for) that have been chugging away for probably 5+ years just running cron jobs, backups, Syncthing services and whatever I toss on them. Throw an SSD in anything and it just works - power goes out, batteries act as UPS. $100 USD each, “just work”.

Thinkpads have always been a Linux favorite, at least the old models when IBM owned the brand but not too sure about the Lenovo modern ones. Last Thinkpad I owned was a 32bit one back in like maybe 2010 and it worked just fine. They tend to be more expensive used than Dells (retain their purchase price better, like a nice used auto).

This is unfortunately a choice the Nautilus (GNOME) folks have taken; in other file managers (Thunar for XFCE, Caja for MATE, etc.) the ability to use custom actions are a first class citizen. Within Nautilus, the nautilus-actions project was superseded by the filemanager-actions project which was then archived: https://gitlab.gnome.org/Archive/filemanager-actions - a custom GNOME action might be something like gio open /path/to/terminal.desktop %d (where %d is the directory from Nautilus)

There are 3rd party attempts to recreate what was stripped out of/abandoned in Nautilus such as this one: https://github.com/bassmanitram/actions-for-nautilus

Went down the rabbit hole for you while drinking some tea listening to the rain - it looks like in the future there is a new app/proposal for FreeDesktop to use xdg-terminal-exec as the new/default way and it’s hard coded into the GNOME “gio” code over here (ctrl+f search xdg-terminal-exec): https://gitlab.gnome.org/GNOME/glib/-/blob/main/gio/gdesktopappinfo.c

That said, it looks like the nautilus-open-terminal Nautilus extension is shipped as part of gnome-terminal so it’s hard coded to run that terminal not using the above code. Instead, you’d need to leverage a different extension called nautilus-open-any-terminal for now until the landscape changes: https://github.com/Stunkymonkey/nautilus-open-any-terminal

(disclaimer: not using GNOME/Nautilus or Fedora, theorycraft from me)

It’s a 4x4 MIMO, most likely LTE+5G; a service like T-Mobile Home Internet uses a hybrid design combining the two (B66 + n41 e.g.)

I’m familiar with the news about the brick - in the past I’ve had this problem (I think it was a bricked… pixel 2?) and faced similar power off issues. Keep trying what you’re trying but in various ways - I vaguely recall that I had to press volume up first and then hold power or something like that (meaning pressing them both at once or power first didn’t work). One of the various combos you’re trying is supposed to be the one that forces it off after ~30secs of holding but a fuzzy memory reminds me it was real finicky to actually get working. Worst case scenario, just let the battery die. :(

To your multiple IMAP concept, I have been using isync / mbsync (name change, package isync in Debian) for years running via cron script to pull email from one domain at one provider and push it to a subfolder of another domain at another provider. You have to be aware of one specific gotcha but it’s otherwise been working all by itself forever without issues. Take note of the PipeLineDepth 1 for IMAP service providers which throttle your speed, I have to use it on the destination side provider config.

Two tips having worked in the corporate world (strict controls):

• Create a basic non-spam web page for it that has something that doesn’t look like SEO garbage or whatever. Nothing more than “hey this is a personal domain of the flatbield family” is fine, maybe a link to something (links enhance rep - put a picture of your dog up or link to a wikipedia article or something) and let it rest for at least 30 days. The 3rd party filtering services used by corporate players severely limit, block or distrust a domain newer than 30 days (or longer, depending). Set up a SSL cert on it for another +1 to it’s rep value, HTTPS is looked at by these services and ensure the CA record is in your DNS for that SSL issuer.

• Ensure you use the Providers’ setup for DKIM, SPF and so forth (many like Fastmail have a DNS-check wizard to get you all set up) as many modern providers will instantly downvote you if anything is missing or wrong with these controls (I’ve heard GMail and O365 particularly). In 2024 these are a must-have, not a nice-to-have, for getting your email received by anyone and everyone.

If you chose a domain at a TLD which has/had been used by the bad buys (dot-xyz, info, zip, etc.) you may wish to reconsider - there are TLDs which are wholescale blocked or downvoted in rep based on this (by the same services used above). Ensure someone working at a bank (strict egress controls for their employees) can visit your domain as a good litmus test as to it’s validity for use in email reputation.

A company such as Fastmail spends a lot of time ensuring their IP address space for sending and receiving mail is clean - getting spammers off their service, getting IP rep cleaned off blacklists and so forth. So your task is to focus on the same thing for your domain - if someone had previously owned the name they could have gotten it on blacklists long ago, a handy way to check old history is looking it up at web.archive.org for captured snapshots (and I’ve walked away from domain names because of this once I discovered previous content I didn’t like).

Fastmail has one feature many others lack (which is hard to research unless you want/need it and have go down the rabbit hole) - scope limited login tokens for specific uses. Specifically, you can set up one for “read only IMAP” (to archive emails using scripts etc.), “SMTP only” (to send emails from scripts like backup reports etc.) and so forth. Many, if not most, other providers either don’t have it, or if they do it’s very limited like one token only with no scope control. $0.02 hth

I would agree, and would bring awareness of ionice into the conversation for the readers - it can help control I/O priority to your block devices in the case of write-heavy workloads, possibly compiler artifacts etc.

The Linux kernel uses the CPU default scheduler, CFS, a mode that tries to be fair to all processes at the same time - both foreground and background - for high throughput. Abstractly think “they never know what you intend to do” so it’s sort of middle of the road as a default - every CPU cycle of every process gets a fair tick of work unless they’ve been intentionally nice’d or whatnot. People who need realtime work (classic use is for audio engineers who need near-zero latency in their hardware inputs like a MIDI sequencer, but also embedded hardware uses realtime a lot) reconfigure their system(s) to that to that need; for desktop-priority users there are ways to alter the CFS scheduler to help maintain desktop responsiveness.

Have a look to Github projects such as this one to learn how and what to tweak - not that you need to necessarily use this but it’s a good point to start understanding how the mojo works and what you can do even on your own with a few sysctl tweaks to get a better desktop experience while your rust code is compiling in the background. https://github.com/igo95862/cfs-zen-tweaks (in this project you’re looking at the set-cfs-zen-tweaks.sh file and what it’s tweaking in /proc so you can get hints on where you research goals should lead - most of these can be set with a sysctl)

There’s a lot to learn about this so I hope this gets you started down the right path on searches for more information to get the exact solution/recipe which works for you.

On my Subscribed view on lemm.ee sorted by Hot which very recently upgraded lemmy-ui to 0.19.4 (your lemmy.sdf.org instance has not upgraded yet, just checked), of 14x news items/links which should have thumbnails - 8x of them are broken/missing, just over 50%.

It’s possible what you are seeing is more visible than before due to issue #2433 which is:

When images are broken, a blank space is all that is shown. A fallback image for broken images would make this more apparent.

With the 0.19.4 update I notice it a lot more now and see a lot from apnews.com causing this, on my Subscribed view with [email protected] and [email protected] (and others) this morning I count:

• 3x missing from apnews.com
• 2x missing from www.theguardian.com
• 1x missing from www.aljazeera.com
• 2x “link-icon only” to www.reuters.com

Sites that are fine: www.cnbc.com, www.nbcnews.com, www.bbc.com, www.cbsnews.com, newrepublic.com and a bunch of fediverse instances and other “not mainstream” sites from around the internet. it would appear that whatever has changed in code is interacting badly with certain mainstream news sites and the ability to get a thumbnail - I browsed the lemmy-ui commits and a metric ton of 3rd party dependencies were updated, gave up looking for the cause.