@Seirdy@pleroma.envs.net avatar

Seirdy

@Seirdy@pleroma.envs.net

Skim before following: https://seirdy.one/about/fediverse-greeting/. It describes how I accept follow requests, block people, etc.

Interested in #Accessibility, #Privacy, #Security (in that order).

I am made of microplastics and can be trusted with your forklift. @alizasystem's boywife, together since 2023-12-04.

tech-stuff: check my "uses" page: https://seirdy.one/about/uses/
Other tech interests in no particular order: linked data, the #IndieWeb, the #Gemini protocol (more into the community than the technology).

Politics: Leftist, capitalism bad, anti-consumerism.

Neuro-atypicality: #anxiety, #ADHD, #ActuallyAutistic.

:QueerCat_Pansexual:

Boundaries: if you're a minor or if we've not meaningfully interacted before, then don't be lewd/flirty with me. otherwise it's fine, in moderation.

Hashtags for #fedi22 searchability: #shitposting #poggies #LinkedData #SemanticWeb #panro #InclusiveDesign #ScreenReader #SearchEngines #anime #webdev #blogging #linux #Fedora #Sway #zsh #IndianAmerican #StarTrek

[Verifying my OpenPGP key: openpgp4fpr:AC6AF1F838DF3DCC2E47A6CF1E892DB2A5F84479]

This profile is from a federated server and may be incomplete. Browse more on the original instance.

brook, to random

Since the beginning of this year I've been maintaining a Mastodon server blocklist called Garden Fence, available at https://gardenfence.github.io/

As the Discord admin space that was being used to coordinate with other admins on the list has since closed, I've set up the account @gardenfence to pick up communication and notifications around the blocklist.

In particular I hope to expand the list of volunteer reference servers in the near future to replace a server that plans to leave the list, to keep the list at a reasonable size.

#GardenFence #Blocklist #FediBlockMeta #MastoAdmin

Seirdy,
@Seirdy@pleroma.envs.net avatar

@brook @gardenfence DM’d a few suggested sources for the first two categories.

I would also encourage you to compare your output with other consensus lists like mine, depending on whether you want to block more or less than a given list. I’m guessing you wish to block a more than FediNuke but IDK if you want to block more or less than my tier-0 list.

hrefna, to fediverse
@hrefna@hachyderm.io avatar

No, seriously, what the hell am I supposed to do with this object when it can literally be any of

  • A URI
  • An object
  • A list of either of those
  • A list that contains both of those.
  • An object of type collection which then contains either/both

Like…

just…

even a peripheral attempt here requires either/both just chucking a bunch of it out of the window and declaring an actual object model or building a billion custom serializers.

/rant

Seirdy,
@Seirdy@pleroma.envs.net avatar

@hrefna …maybe it’s just an elaborate troll?

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

The three most popular DNS protocols with transit encryption are DNS-over-HTTPS (DoH), DNS-over-TLS (DoT), and DNS-over-QUIC (DoQ). This should help you choose what to use:

  1. Do you actually need to override OS DNS support? If not, or if you’re unsure, go to 6.
  2. Are you ready to implement DNS protocols correctly, or add a dependency that does so? If you’re not, go to 5.
  3. Does the network filter DNS traffic? If it does, go to 5.
  4. Do you already have QUIC support? If not, use DoT. If you do, use DoQ.
  5. Do you have an HTTPS stack? If you do, use DoH.
  6. Give up and delegate to the OS.

Let your HTTPS stack handle HTTP/1.1 vs. HTTP/2 vs. HTTP/3 support; don’t treat DNS-over-HTTP/3 as a separate protocol. I don’t know enough about DNSCrypt to make an informed recommendation about it, but DoQ and DoH meet my needs well enough.


Originally posted on https://seirdy.one/notes/2023/11/18/choosing-an-encrypted-dns-proto/ (POSSE).

astrid, to random
@astrid@fedi.astrid.tech avatar

a fate worse than Palo Alto

Seirdy,
@Seirdy@pleroma.envs.net avatar

@astrid Los Altos

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

What’s a good actively-maintained CLI (not TUI) TOTP/HOTP two-factor authenticator? I want 2fa for some sysadmin tasks (like ssh, connecting to certain endpoints, etc) that doesn’t involve my phone or key fob so i can get something resembling per-device approval on a subset of devices with my keys installed.

Sparky, to random

mfm parser fucking sucks

Seirdy,
@Seirdy@pleroma.envs.net avatar

@emmy @Sparky and it is a nightmare to parse. So much backtracing. it’s actually a very complex superset of HTML.

Seirdy, to webdev
@Seirdy@pleroma.envs.net avatar

Apparently the venerable HTML-Tidy is now completely incompatible with multi-line .

Reverting to the commit before the one causing the regression also reverts a lot of important fixes that depend on it. That commit was the final step in enabling a new non-recursive parser.

So I guess I’m dropping HTML-Tidy from the seirdy.one build pipeline, after less than a day. Upstream development appears to have stagnated.

Anybody else know of a good XHTML formatter?

Seirdy, to webdev
@Seirdy@pleroma.envs.net avatar

I have an question:

Is there an agreed-upon way to specify more than one generator in <meta name="generator">? My site uses Hugo for building, but also uses (a personal fork of) HTML-Tidy and xmllint for post-processing. The former offers its own <meta name="generator"> field.

I disabled the HTML-Tidy generator tag so it wouldn’t conflict with the Hugo generator tag, but I’d like to keep them both if possible. Would I simply add multiple meta elements, or use a comma-separated list in the content attribute’s value?

(Before someone asks: my rationale for including one at all is that web surfing projects can allow people to find sites based on technologies used. Marginalia Search, for instance, allows filtering results to a given generator.)

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

Reply to We Need to Bring Back Webrings by Arne Bahlo

Webrings are alive and well; they don’t need to be “brought back” because they’re already here.

I’m in 14 webrings. If you think that’s a lot, foreverkeith.is has a growing list of 100 or so.

Some have gotten quite large. The Hotline Webring is close to 700 members. The Yesterweb Webring recently shut down because it got too large, past 800 members! Some of the people behind the larger Yesterweb community felt that the webring had gone too mainstream and become a venue for SEO and traffic-boosting rather than a chill network of friendly like-minded people.

At this point, webrings aren’t some dated concept that’s become tiny and niche. They’re quite mainstream outside of the commercial web.


Originally posted on https://seirdy.one/notes/2023/11/14/webrings-are-already-back/ (POSSE).

mariyadelano, to SEO
@mariyadelano@hachyderm.io avatar

The piece I’m working on right now has taken me 60+ hours and counting.

And the hardest part, by far, has nothing to do with finding the time or the energy to do this - it is the immense sense of responsibility that I feel to get this RIGHT. Because most for the things I write under my name are bigger than me.

I’m writing to fight for what I believe in. For the internet. For my industry. For good ideas. For good people who want to do good in the world.

And today I fight for all of

Seirdy,
@Seirdy@pleroma.envs.net avatar

@mariyadelano you might be interested in the idea i put forth in this poll: https://pleroma.envs.net/notice/AbgSNTL5DQB6O6iYd6

carving out a subset of SEO focused on being compatible with and friendly towards scrapers, such as search engines + removing the bits around gaming the algorithm and outranking your competitors in a race to the top that leaves hobbyists and noncommercial sites at the bottom.

Seirdy,
@Seirdy@pleroma.envs.net avatar

@mariyadelano thanks!

you might also find this helpful: https://seirdy.one/posts/2021/03/10/search-engines-with-own-indexes/

Engines like Marginalia, Stract, and Kagi’s “noncommercial” lens remain much less impacted by aggressive SEO by deliberately downranking commercial sites. I think it’s important for the noncommercial Web to have an area where it’s still accessible, and not drowned out. It’s a nice glimpse at what remains of the old Web before adtech and SEO grew dominant: people just sharing what they know.

Seirdy,
@Seirdy@pleroma.envs.net avatar

@mariyadelano if you want someone to look at a draft before you publish, you can always reach out. :gura_wink:

mariyadelano, to random
@mariyadelano@hachyderm.io avatar

Finalizing my article responding to the Verge's SEO piece.

Current working title:

Seirdy,
@Seirdy@pleroma.envs.net avatar

@mariyadelano @liztai I feel like most people disappointed about SEO spam have never used a search engine besides Google, Bing, or a Bing-based engine (e.g. DDG, Yahoo).

Seirdy, to Blog
@Seirdy@pleroma.envs.net avatar

New post just dropped! Proposal: an HTML element for spoilers. The case for including dedicated elements for spoiler tags in HTML. An informal proposal describing use-cases, syntax, semantics, and recommended user-agent behavior. Here’s an excerpt:

While browsing the WICG discussion forms, I stumbled upon a proposal for a standardized . I made two comments, and stopped myself before writing a third; this called for a blog post. I think HTML should have a spoiler element.

RE: https://akko.wtf/objects/8fd2f79b-9529-441d-8b6f-079ac4a8f599

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

seirdy

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

I am once again begging everyone to understand that verified boot, enforcing app signatures, TPM, etc. are actually good if they protect the user. In the status quo, they tend to protect the vendor from the user.

If the user defines what’s allowed to run, then there’s no problem.

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

yo who wants to review a draft blog post of mine that proposes adding a spoiler element to HTML?

if you’re interested then LMK your preferred reading format (HTML, GemText, or markdown filled with custom Hugo shortcodes) and I’ll DM you a draft. image links may be broken but there’s alt-text.

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

What’s a good self-hosted fulltext search engine setup that ticks these boxes?

  • Fine for small indexes with under 1k pages.
  • Lightweight, using under 150mb RAM (ideally much less).
  • Can be built with a C, C++, Go, or Rust toolchain.

I’m looking at building something with Typesense but it looks way overkill tbh. I like the approach of Pagefind, which builds an index all at once when you build a static site. I don’t want the search to happen client-side, though. No client-side JS/wasm.

Seirdy,
@Seirdy@pleroma.envs.net avatar

Right now, I’m looking at Sonic. It has a really simple TCP protocol for communicating with the server. I’ll probably write my own client that I can talk to remotely…or just use SSH :02teehee:

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

@packetcat does anybody know how dnscrypt-proxy compares to encrypted-dns-server + doh-server, also by the same org? Besides the language used.

Then there’s dnsproxy, which also offers DoQ alongside DoT and DoH. DoQ is marginally faster than DoH3, but also easier to filter.

Both DoH3 and DoQ solve specific problems I have: I need transit encryption, deal with high-latency lossy connections, and frequently experience connection migration.

RE: https://tenforward.social/users/packetcat/statuses/111371384318733326

18+ Seirdy, to random
@Seirdy@pleroma.envs.net avatar

I was again asked about including an instance in my blocklists based solely on its associations. I explained why I didn’t do this in my documentation:

Finally, you could just do something else entirely. I never use “your blocklist is different from mine” as a block-reason; even “totally open federation” is never the sole reason for pleroma.envs.net suspending an instance. Suspending instances for not using my own blocklist feels wrong; it’d turn these lists into a source of authority and cross the line from moderation to cop behavior.

I’m providing a resource but I’m not enforcing its usage. That means not having a list of instances you must defederate from, lest you be defederated yourself. In other words: making a blocklist disqualifies me from using federation alone as a block-reason. I’m not trying to use my lists as a tool to build authority.

You might be comfortable with the idea, and I won’t stop you. This is why I can’t share that comfort. Instances that are really cozy with blatant Nazis n shit usually have other reasons to block. I cite those.

Seirdy,
@Seirdy@pleroma.envs.net avatar

I accidentally made a top-level post instead of a reply to this, so I’m repeating it here:

You’re welcome to use, e.g., “federates closely with a member of FediNuke” as a reason to block on your own instance if your members are cool with that. Maybe message another admin you’re friends with if they have similar policies. I’m not policing your own moderation policies.

Beyond that, I don’t condone globally making block recommendations solely on this basis.

Trust me: for the worst associations (e.g. NiceCrew, DetroitRiotCity, ClubCyberia, et al) there’s almost always a more clear-cut reason.

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

bottom.

Seirdy, to random
@Seirdy@pleroma.envs.net avatar

Daily reminder to everyone finding SEO garbage on Google for short-tail searches that Marginalia, Stract, and others exist.

Also: this isn’t a Google/Bing problem, it’s a Web problem. If you build a new search engine with naive solutions to crawling, indexing, and ranking, then you’ll find SERPs full of SEO spam too. Google’s particularly bad at rewarding listicles, but actually has to do a ton of work to downrank the more blatant spam that its system would otherwise reward.

Seirdy,
@Seirdy@pleroma.envs.net avatar

Oh and anyone who markets their search engine as “unbiased” is lying. An unbiased engine is just biased towards whoever has the biggest SEO budget.

Seirdy,
@Seirdy@pleroma.envs.net avatar

TBH, I kind of want engines to have bias. Downrank inaccessible, privacy-hostile, scam-peddling sites. Reward good behavior.

Seirdy, to github
@Seirdy@pleroma.envs.net avatar

search got a redesign. Here’s a screenshot of the searchbox for logged-in users with forced-colors enabled (Firefox on Linux).

The text is invisible. I have to highlight it or look at the top suggestion to see what it is. When I submit the search, the search box looks empty; I have to switch to the search box and highlight the text to see what I searched, or eyeball the query parameters in the URL bar.

This is why I’m still using Firefox’ dedicated search box instead of the unified URL/search bar; I have a keyword shortcut for searching GitHub.

I should make a counter for days since a UI got a redesign that significantly hurt accessibility. I can’t remember ever seeing a redesign that made things better. I’m quote-posting the last time I made a post about this happening.


RE: https://pleroma.envs.net/objects/d46827df-3535-422c-add7-2206f6a17297

Seirdy,
@Seirdy@pleroma.envs.net avatar

The unusable search bar and link colors that circumvent forced colors aren’t the worst of it.

The worst part is invisible carets and highlights in the code viewer. How am I supposed to select and copy the right text when I can’t see what I’m selecting and copying?

Breaks keyboard nav too. I have to open the raw file, and at that point I might as well curl it directly into my $EDITOR and sidestep the browser altogether.

When your web UI has regressed to become worse than curl and you’re investing in AI and making it even worse, you need to fucking fire some higher-ups.


  • All
  • Subscribed
  • Moderated
  • Favorites
  • megavids
  • rosin
  • ngwrru68w68
  • osvaldo12
  • DreamBathrooms
  • mdbf
  • magazineikmin
  • thenastyranch
  • Youngstown
  • khanakhh
  • everett
  • slotface
  • tacticalgear
  • kavyap
  • JUstTest
  • normalnudes
  • Leos
  • GTA5RPClips
  • ethstaker
  • InstantRegret
  • cubers
  • modclub
  • Durango
  • provamag3
  • cisconetworking
  • tester
  • anitta
  • lostlight
  • All magazines