On the Pi I already use an HDD for data and most of my programs, plus some megabytes (64 today, yesterday it was 32) of swap in addition to the 128 MB I have on my slow microSD, (which apart from that only holds the base system, really).
Considering that an SSD on the Pi3 is wasteful due to the USB 2.0 limit, that I don’t have any spare, and that the smallest SSD I can find (128 GB) is still at least twice the cost of my 320 GB HDD, guess what 2.5" external drive I’ll keep as main currently.

System and programs aren’t running slow right now, so, since I need more swap, I could try adding some more on the HDD, and only then actually try pendrives, just for 1 o 2 GB of swap at max. I can add them, since I have some spare and they are cheap to get anyways.

But yeah, if I had USB 3.0 on a SBC with 1 GB of RAM, I would get the SSD.

Flashdrives die sooner than good quality SSDs, but in general are at least a bit more resilient than microSD cards. Even I’m considering starting to use a small USB pen as extra swap space for my Raspberry Pi 3; for what they cost, even if they will eventually break, they are good value.

Why is it slow for you? Here are some things to consider:

• USB speeds: USB as a protocol is generally slower than SATA, even version 3. If your drive or your port are 2.0, then you will deal with a theoretical max bandwidth of 480 mbps (but the practical one is much lower!). If they are 3.x, check that the flashdrive makes proper contact in the port with not only the main big 4 pins, but also the small pins hidden deeper inside the connector (I have no idea how on Linux tho :/); most USB A 3.0 connectors and ports have issues connecting together for some reason, and it may happen that the SuperSpeed pins can’t get close and thus the connection will degrade to 2.0 speeds.

• Cheap chips (pun intended): Cheap USB pen drives nowadays use both low-quality controller chips and memory chips. From experience I can say that some inexpensive modern flashdrives that are sold as USB 3.0, and get connected as 3.0 to an host system, actually work at around 2.0 speeds. Why? Sometimes it’s because low-quality memory chips are used, that won’t work properly if accessed with high speeds, and thus the controller chips in the drives are programmed to limit read/write speeds… some other times, the controller chip is also crappy, and transfers slow down because it overheats. Also, bandwidth is not the only thing that suffers: on cheap flashdrives you can experience high latency.

Both limited bandwidth and high latency impact a lot when you use any storage medium for OS or program storage, or even swap memory. If you want some decent but easy to read statistics to test those values (with graphs too), install GNOME Disks and use its disk benchmark tool. You can verify how well your USB pen drive performs, and even compare it to your SSD.

As a side note, from most of my flashdrives (including one which connects via USB 3.0) I always get no more than 15 MB/s read + 5 MB/s write… that’s a practical speed of 160 mbps, a mere 1/3 of the theoretical for 2.0.

comments aren’t a thing with writefreely

What do you mean? You can look up any profile in the form of @blogname@writefreelyinstance.domain from apps like Mastodon, Friendica, etc…, see all posts, and comment regularly.
Is the fact that there’s no link or embed of the comment section at the bottom of the WriteFreely page that is bothering you, or am I not understanding?

On a side note: if you are really choosing how to build a blog (like it seems you are), and are not taking the first free managed hosting provider you come across…
I would think twice before using any server software instead of keeping your site static. Having a server software that’s more complex than simply serving static files will do more harm than good in the long run: more security flaws, you have to always keep the thing updated, higher resource usage, and hard to make your content survive the test of time (backing these things up is hard and when you do, you have a database file, not some plaintext ones)

You gotta hate people that are different from you a lot, to be able to go as far as making a videogame mod like this, wtf

Miis actually look good compared to this, because their design is simple, kind of chibi, and doesn’t try at being too realistic or accurate

This is in the uncanny valley, Miis really aren’t

I’m glad a part of my devices still are micro USB-B…

I, on the other hand, can’t understand why, in a big part of the normie Internet sphere, the Metaverse has been hyped and appreciated so much - by the same people that don’t know Mozilla had basically already created the Metaverse, many years prior, with Mozilla Hubs.

Friendica has a follow thread feature and yes, it’s so convenient. Lemmy should have that as well imo.

I actually do know that… but it still has higher probability of convincing someone, compared to privacy

That’s actually sadly real, and I can’t emphasize “sadly” enough 😅

I have a friend that about 3 years ago introduced me to libre software. At that time, she used very few proprietary garbage. From about a year though she went back to using Instagram (why???) and other stuff (not only social networks, but things like messaging apps, which is a bit less bad), but luckily still prefers libre software when possible.

I still have to thank her for being so insistent at the time about “you have to remove google play services!!!” and other stuff. Even if she is now cringe, and not a based free software purist anymore, it’s partly thanks to her that I’m now a more aware computer user.

Well, Friendica is libre software and a federated social network. Not highly controlled.

As with plagued with uselessness… Actually, doesn’t differ from the real world. Depends on who/what you like to see (at least on fedi platforms, where you choose, not a heuristic algorithm).

Put it on ethics and mental health: say how proprietary platforms don’t give a damn about any of their users, and won’t think twice about abusing them (with vendor lock-in, ads, and, most importantly, causing addiction just to make you use them more so they can milk you for money)

Also, maybe explain how moderation on there is an absolute turd, with platforms often punishing users which break their ToS less, if those users make a lot of money to the platform (think influencers); unlike on decentralized platforms, where you can choose an instance where the rules satifsy you and you see the moderators working good.

Nah, let’s tell them about Friendica and they will be happy

Unlike Chromium though, Firefox still lets you theme a lot of stuff, including position and dimension of UI elements, via browser-level CSS.

Online, you can find many themes for many work styles, or you can make your own. I want my browser to be compact and minimalist, for example, so I made this: https://mastodon.uno/@octo/108329542312201239

You are telling me to do this? Me?

I am the same person who the other day created a wiki about a really niche matter, and will also need to make a page about myself because the main subject the wiki exists solely because of me… 😆

Seriously btw, usually frequenting social spaces (online or offline, doesn’t matter) you leak stuff that’s way more important than the gramatical gender you use or want other to use to refer to you. You leak what are your interests, as well as other unerasable personality traits. I even wrote a blog article about this.

Today I basically only look for lossless music, which complicates things, since I can’t just use yt-dlp and solve my problem.

The few times I can buy lossless music, i do so on Bandcamp, paying with PayPal.

When I can’t pay, my order is the following:

• Bandcamp (nice artists often put some free stuff too on there)
• Archive.org
• Going on adventure on search engines trying to find FLAC torrents, reupload websites or (it actually happened one time) unprotected private HTTP servers (!) that Google will just publicly index, absolutely not giving a damn about the fact that the server exists just because someone forgot to close a port on their router

This is one of the reasons why I value my data to be public. There are some things that I unfortunately consider sensitive and so don’t want to make public, but many other things are shareable for me, and so I make sure to share them around.

For example: my git repos. The important ones are all public, and mirrored on the servers of many providers. If I somehow lose all my passwords (and it shouldn’t happen, considering that I have a good backup plan for my password database file)… a good part of my data is public, I can just download it, then restore it to some new accounts.

Same goes for things I like to write (my blog for example is public and mirrored), some photos and videos (but not with me or other private people as the subject), maybe memes, minor thoughts, and things like that (my Mastodon profile is public)… and that’s pretty much it.

And this strategy, for me, can be even more calming if I not only accept that other people might archive or reshare my data, but even encourage it (by clearly marking things as available under a libre license).

Anyways, for my sensitive/private data (including my password db) I have a complex backup strategy, too long to share in detail here for now.

Am I worrying too much about losing data? xD

we don’t call it “piracy” here, we call it “sharing free of charge for the greater good”

As far as I understand, AMD has an equivalent technology, called PSP.

The difference is that at least AMD let’s you disable that, by allowing motherboard manufacturers to distribute BIOS binaries without PSP, that the user can download and install… But this is up to the motherboard OEMs 😕

With a good (laser) printer you could create your own with open (but general-purpose) formats: QR codes or (my favorite) Twibright Optar codes, encoding music with any audio codec you prefer (Opus would be better). Then you can make a machine reading your printed paper strips as fast as possibile for high quality realtime audio decoding

Protip: You can effectively get rid of Google Services on whatever is your current phone, without even needing root, let alone installing a custom ROM.
Check out this guide: https://www.xda-developers.com/uninstall-carrier-oem-bloatware-without-root-access/

Keep in mind though: if your phone uses Google apps for Dialer, Messages, and Calendar (and you need those features because most likely you want your phone to be a phone, not a mini-tablet), you will also need to uninstall those app and install full replacements.
I’m personally going with: QKSMS for messages, and Lineage OS dialer and calendar (APKs extracted from a Lineage OS ROM for a random phone and the same Android version as my phone)

I know, that’s why I put emphasis on the encryption!

Protip: if you need cloud storage for backup purposes (you shouldn’t store your files primarily on a machine you don’t own, but backup is another story), use Rclone. It’s a sync client for many cloud storage providers, but has amazing features.

If setup correctly, it can encrypt your data locally before uploading it to the cloud, and can even make a single seamless big cloud drive from joining together many small (free :D) accounts.

It works well, I have it automatically sync my stuff encrypted to the cloud from my NAS every night.

They can’t do much if you use any proxy, be it Tor, a public single-node proxy, or a commercial VPN (which I think is inadequate for an use like this, after all VPN providers are ISPs in a way, so you are just hiding traffic from your own ISP and giving it to another one).

They could just see the traffic encrypted between proxy client and server, and maybe do some correlation with time of usage and stuff like that.

If you have bad opsec, it’s possible to completely deanonymize you on Tor and track what you do, but it’s not easy and so it’s something that only law enforcement does. Entities with simple commercial interests don’t go above certain limits of complexity. Most websites for example use browser fingerprinting techniques, and as far as I know some stuff uses advanced user fingerprinting tactics (logging the unique way you move the mouse or type on the keyboad, like Google ReCaptcha 2 does), an ISP could do even less.

Depends.

First, the content of HTTPS traffic might be encrypted, but some metadata (the domain name or IP address of the destination, for example) is not. You can use just that to do some tracking.

Then, if for example the ISP sees an user visits a particular website with poor privacy practices (let’s say, any social media run by big tech), they could do business with them and buy and sell data.

Lastly, but this is tinfoil-hat territory, what is to say they can’t leverage the biggest flaw in the global HTTPS infrastructure: they could manage to corrupt a certificate authority into giving them their root certificate keys (or at least into decrypting traffic they start logging and then pass over). This way, the ISP could read the traffic of any user of any service which uses certificates emitted by the corrupted CA. Very difficult that something like this happens because big CAs aren’t just ran by 1 omnipotent person, but technically 100% possible.

Side node: Tor solves the trust problem of the global certificate infrastructure: Hidden Services don’t work with trust on a CA, they work with just a public and private key system, where the only weak point is the server of the hidden service, storing the private key it uses to encrypt traffic and serve it to clients.
Edit: the server’s private key decrypts data the client encrypted with the server’s public key, and viceversa, idk why I always mess the 2 things up in my mind :P, but the point remains

Wrong. It’s “basé et rougepillé”

Time to start using Tor for everything I suppose… (Or just hope this tracking doesn’t come to Italy too)

The new GNOME text editor is cool but lacks many features I need

My trial of it went like this:
> Upgraded to Fedora 36
> “wow new text editor?”
> See it lacks features like extensions, also the option to use tabs instead of spaces doesn’t seem to work
> Uninstall it and install gedit…

I hope the new editor catches up to gedit ASAP, because I like it aesthetically

The PinePhone is probably the only one I could ever buy, because all the other ones (unless there’s some new out there) are too costly, both phones that come without Android, and Android phones for which the community ported Linux (no one ports good ROMs for cheap phones).

But, at that point, since the experience isn’t really that smooth for a smartphone (never tried it, but the majority of people who did say so), might as will just build myself an UMPC with a cheap Linux SBC, and also have perks like a built-in phisical keyboard. Then I would just use a featurephone for calling, SMS, taking pictures, and WiFi hotspot for my UMPC.

Another good tip I would suggest (since I actually use it), is to keep JavaScript disabled by default in your browser.

When visiting websites, if one is broken, I enable JS just for it. If it’s just a static site (a blog, news article, …), and is developed correctly (there are news sites which don’t work at all without JS, I wonder how can one mess up so badly in website development), then keeping JavaScript off will not alter the site’s real functionality, but will block ALL bad scripts, and also make the browsing experience faster.

The future looks grim. What smartphone am I supposed to have in the coming years? Even if my current one never breaks beyond repair, at a certain points all the apps I need will have become too slow or even incompatible. Personally, I can’t just use a 10 year-old version of my favourite web browser, messaging app, or the likes, and I need at least the basic freedom of having root privileges.

And by the way, this doesn’t only apply to smartphones, the situation has far worse implications. The PC transition from x86 to ARM will happen, slowly but it will happen. ARM has countless benefits which I advocate for, but the fact that it is a customizable architecture (there are countless ARM chip producers, every one of them making chips that are incompatible with competitor ones), mixed with the fact that OEMs don’t have to release sources for their SoC-specific blobs, is a recipe for the disaster of closed platforms.

You will be stuck using Windows on your shiny ARM “PC” with locked UEFI bootloader (and even if those remain unlocked, someone will have to spend time and effort to port other OSes to each specific computer, like you see today on smartphones with custom ROMs) and you WILL be happy. Unless you want to spend an high price on some low-end hardware sold by companies that unfortunately, to survive with selling open hardware, have to come to this compromise.

Honestly I’ll disagree.

I think the biggest problems of slowness actually come from matrix.org. When I switched to my selfhosted homeserver (hosted on a Nintendo Switch in my house, no fancy costly VPS) my whole experience got faster and better.

Talking about clients, there are some promising ones out there but which are not ready yet. Element and FluffyChat are really the only unique ones (not counting forks) I would define as complete.

Personally, I have been using SchildiChat (Element fork) on Android and Element on desktop for all my chatting in the last few months, and I mean all: since I started hosting my homeserver, I also set up puppet bridges for Telegram and Discord, which means I can now chat in any room from these platforms, on my Matrix, seamlessly.

Element for both platforms is maybe overall slower than Telegram, but is onestly much more stable and better performing (speed isn’t the whole of performance). Before going full-time-Matrix, I have been a full-time Telegram user for about 4 years straight, so I think I’ve seen how things are.

• Element desktop doesn’t suffer from the strange memory leak and performance drop problems (occasional freezes, or high disk usage) that Telegram desktop does. Element just sits there at ~150MB, Telegram can go up to ~1GB+ after some hours (if you have many chats at least)
• SchildiChat (Android) somehow allocates ~400MB of RAM (like the stock Element for Android) (wtf?) but is stable, it doesn’t have the infinite small bugs and problems that Telegram for Android has. Also, SchildiChat doesn’t drain my battery. For about 2 hours of foreground use, it uses up ~50mAh, while Telegram (doesn’t matter if stock or a fork like Telegram FOSS or Nekogram X) wastes iirc ~300mAh.

TLDR: I’m never looking back from Matrix.

Side note: have you seen the source code for Telegram for Android? No wonders it has all these problems… Apart from being just bad, it’s also developed in a sus way. Telegram desktop is developed like you would expect any big open source project, with frequent small commits created by different people. Telegram for Android has 1 big commit every few months or so created by always the same person.

I never thought that holders of proprietary platforms could come close to this level of influence, making the developers of a libre library have to relicense their work, or otherwise app developers can’t publish for those proprietary platforms if they decide to use the library in question.

I’m a strong (A)GPL fan but I see why proprietary companies hate it, but what is wrong to them with the already permissive LGPL?