thayer

joined 1 year ago
[–] [email protected] 11 points 3 months ago

I've been very happy with both Silverblue and Kinoite. I've installed it to all of my workstations now and can't imagine ever going back to a traditional distro.

Your comments suggest that you're already aware of distros like Silverblue so, if I may ask, how are these different than what you're looking for? Silverblue comes with several flatpaks installed, but you can easily remove these and you'll be left with a pretty barebones ostree image.

[–] [email protected] 2 points 3 months ago* (last edited 3 months ago)

I've been super happy with my 8th gen Intel NUC i5. I put it in an Akasa Turing fanless case, installed an NVMe for host OS, and an 8TB SSD for data. It's low power and so quiet that I couldn't imagine ever using fans again.
I also have a USB 3.2 drive dock for external backup HDDs, but I only turn it on when actively doing a monthly backup.

8TB holds more media than I'll ever need, but I do trim movies and shows regularly. For some, 8TB won't be anywhere near enough, and SSDs exceeding this are ridiculously expensive.

[–] [email protected] 4 points 3 months ago* (last edited 3 months ago)

We swapped our 2080s for 6800 XTs last year and couldn't be happier...a 7900 GRE should be great for gaming, but I can't speak to LLM performance.

[–] [email protected] 2 points 3 months ago* (last edited 3 months ago)

I was able to extract the img from the ISO using geteltorito as described in section 5 of this ArchWiki article. Once you mount the resulting img file, you'll end up with the same file contents achieved by running their Windows BIOS Utility through wine.

The relevant binaries appear to be under the folders, N24ET76P and N24ET76W. Both scan clean for me, for whatever that's worth:

curl -X POST -F'file=@N24ET76P/$0AN2400.FL1' https://pk.fail
{"details":{"analysis-time":"1.395106993s","hashes":{"md5":"ba73792a5fc831ca84b4cd3a21c03247","sha1":"24a5bb42d670c7705aed06588f0092ec11a32564","sha256":"b9510c73657460ae24c550b71d217a543b0fc3c30a3e081eff31d9d8f1a2bdda","sha512":"8ef6f0dcffbca05b79710b8599b1b1c926ee59185a675bc7eeede6da040c751097303ada523611271de6aaf190a597cdd6e9d5cf564d06987abcf712f61227c6"}},"status":"not-vulnerable"}

curl -X POST -F'file=@N24ET76W/$0AN2400.FL1' https://pk.fail
{"details":{"analysis-time":"1.438471526s","hashes":{"md5":"de1551b0bcc73e19375f7111def72278","sha1":"cd41f36d018f940c308a7be25a20e81bdb7e4cf2","sha256":"b3f646095e47bb94f04390c756cb4133201b1231a8b224174f10bb06bd3835f2","sha512":"55143f4903f92d88057bc9d4232b0d328e9ace36330f35fafdf0485d8bebb3f79b9fedc88ab1dec7fc04a8a3e0890887c1dd7632a2ffa397fb0917be90e3f93f"}},"status":"not-vulnerable"}

The linux command mentioned in the Ars Technica article elsewhere here is efi-readvar -v PK. For Fedora and Arch users, efi-readvar is available in the efitools package.

Edit: Clarity

[–] [email protected] 3 points 3 months ago* (last edited 3 months ago) (3 children)

An ISO is just another archive format, similar to zip, tarballs, and rar files. Most modern archive tools can open and/or extract its contents like any other archive.

Edit:

It looks like Lenovo releases their ISOs formatted as a CD image. See: https://workaround.org/article/updating-the-bios-on-lenovo-laptops-from-linux-using-a-usb-flash-stick/

[–] [email protected] 2 points 3 months ago* (last edited 3 months ago) (5 children)

I can't say for certain, but I think you just have to grab the last firmware binary released for your T480 from the Lenovo website and run it through the online validator: https://pk.fail/

[–] [email protected] 4 points 3 months ago* (last edited 3 months ago)

If ambient noise is a concern, I'd go with an SSD. If money is tight, an HDD will give you the best value.

My server is in an otherwise quiet home office/sitting room, so I went with an 8TB SSD (870 QVO). Spinning disks make a fair bit of noise just waking up, let alone the actual file operations.

[–] [email protected] 1 points 3 months ago* (last edited 3 months ago) (2 children)

You should be able to layer the xdg-desktop-portal-gnome package, which will also pull any dependencies.

To answer your general question though, yes I believe you can easily install at least minimal versions of each DE with little impact to rpm-ostree performance. They don't need to be separate images, though that's possible too by rebasing and pinning. I would just layer the necessary packages to load a GNOME environment (start with rpm-ostree install gnome-shell). This way everything stays up to date with the active image. For example, I'm running GDM under Kinoite simply because I was having unresolvable issues with SDDM and LightDM.

Pinning separate images would require you to rebase with each image update and then unpin/pin the old/new images...too much work.

[–] [email protected] 1 points 4 months ago

Haha, oh I know and I'm all for trying things for the fun of it! Just wondered if there was a practical benefit of such a setup.

[–] [email protected] 1 points 4 months ago (2 children)

Interesting endeavor...any practical benefits? I would think that even a slow USB 2.0 drive would provide better performance than a cloud-based file system.

[–] [email protected] 1 points 4 months ago

Craigslist is still superior in the PNW, at least in my experience.

-12
submitted 1 year ago* (last edited 1 year ago) by [email protected] to c/[email protected]
 

If Jerboa is the official mobile app for Lemmy, why not simply call it Lemmy, or Lemmy Mobile, or Lemmy for Android? It seems more practical to make use of the official platform name, branding and trademark.

Edit: Thanks everyone for the correction, I was under the impression that Jerboa was the official app because it's made by one of the two core Lemmy devs.

 

I'm hoping to start a dialogue about the current use of descriptive community names, usernames, lack of semantic URLs, and other usability issues on the Lemmy platform. I say dialogue because I am new to Lemmy and I can appreciate that some things may be done differently here for specific reasons.

This is not my sandbox but I'd like to see a castle here someday nonetheless, and I'm willing to help make that happen if there's an appetite to see these core issues addressed:

1. Community display names

Community display names should not be used in any meaningful way on the platform. They should not be displayed in the feeds, nor in community search results. Personally, I don't think they should be used anywhere except as a byline on the community's feed.

These descriptive names are not unique and it is trivial for anyone to create a community and change its display name to match that of another on the same instance. This clone will appear right alongside the legitimate community in search results, page feeds, and even moderator lists (such as those viewed on profile pages).

Many community display names are so long that they are truncated when viewed in mobile apps, adding to the ambiguity.

It is also impossible to know the actual community name until a link is hovered over (impossible on mobile) or actually visited (potentially dangerous).

2. User display names

Similar to community names, these descriptive names should not be used in any meaningful way on the platform. There is nothing to stop me from changing my display name to that of a site admin, moderator, or user, and then creating posts under the guise of that person. Again, there is no way of knowing the actual username without hovering over the link (impossible on mobile), or visiting the user's profile directly.

Another side effect of showing display names in the feed is that some users have emojis in their names, or emojis as their name, which is distracting at best and annoying at worst.

In my opinion, display names should be restricted to a user's profile page, similar to how GitHub implements usernames and full names. Post feeds, search results, and any other meaningful place should display the unique username only.

3. Semantic (clean) URLs

A GitHub issue discussing cleaner URLs has been open since July 2020, which leads me to believe this isn't a priority. I won't list the many reasons why user-friendly, SEO-friendly post slugs are important today, as Wikipedia already has it covered (and with a clean URL). The merits of clean URLs have been written about extensively for more than a decade. The bottom line is that this:

https://lemmy.ml/c/lemmy_support/72hsHD/qol_usability_concerns

...or even this:

https://lemmy.ml/c/lemmy_support/1043897

...reveals a lot more about a link before I click it than this:

https://lemmy.ml/post/1043897

It helps to understand the link destination before you click it, and this is an issue that will only get messier if left unaddressed for too long.

4. Sanitized post titles

Last week, I noticed that users are able to include markdown in their post titles, allowing for `code` syntax highlighting in the title itself. This is a bad practice, prone to abuse in the long run as some users will increasingly try to draw attention to their posts.

5. Link posts don't link to the link

I fully appreciate that Lemmy isn't trying to be a Reddit clone but as a link aggregator platform, I'm surprised that link posts do not actually link to the submitted hyperlink. This contradicts not only Reddit, but other link aggregator services, including Hacker News and Lobsters. Currently, the user has to know to click the thumbnail instead of the post title, or enter into the post and then click the title a second time to visit the submitted link. This is not intuitive.

view more: ‹ prev next ›