Require a specific
user-agent string/URL argument/public username-password
Everyone that does not have that gets nothing. New users get it by clicking an obvious button labeled "use-this".
This requires that users are smarter than scrapers, which i admit is a weakness, but most people at least found the Join button once.

Set the whole site under TOR. Who scrapes there?
The surface web site only has a link to the tor-browser and the onion address.
I would start on TOR anyway and just post the link to wikidot, so I would only need to buy a domain on the surface once everything is set up in The Deep Web. Pointing a server away from an onion service onto a surface web domain should be doable, but i have not looked into it for years.

I have no idea whether this is still usable, but it could be worth a shot: HTTP-Code-Tarpit-Talk implemented with nginx.

What about the Mediawiki Discussion Tool does not work for us?

So far I went through the process of actually setting up a MediaWiki instance on an nginx instance (instead of just messing with crappy local solutions) and frankly the process is quite simple. You just extract the MW provided archive, tell nginx where the root lies, and paste a ~20 lines config and that's pretty much it. However, I'll mention the two issues I ran into for the sake of posteriority anyway:

1. Don't just apt install mediawiki; that package (at least on APT, idk how it is for AUR, flatpack, etc.) comes with an apache2 webserver that, during the APT installation process, is going for full on murderization of your beautiful, lightweight nginx server¹. You can install MariaDB the necessary php dependencies manually, the Installation Requirements give you a handy overview of packages].
2. I initially had problems actually running the wiki software because for some godforsaken reason, it was expecting a php8.1-fpm socket for cgi, whilst I had php8.4 installed. I tried messing around in the nginx config to adjust that path, but after some time I gave up and simply ln php8.4-fpm.sock php8.1-fpm.sock in /var/run/php. That surprisingly worked, even thought I never expected PHP to actually use semantic versioning. Could this cause nigh-undiagnosable errors down the line? Maybe. Probably less than having two concurrent PHP versions installed? Certainly.

So I guess my next step is to try that again on an actually public-reachable server with its own domain. Since nginx also implements a mail server (which you'll need because of domain registration shenenigans) that's something I should look at. Besides that, it'll be development of the wiki theme and dealing with the forum and page ratings.

The latter is actually pretty straightforward now, becaus by virtue of being a lazy bum it seems like the RatePage extension has been developed to include basically every feature we would have needed in the meantime.

Forum… eh. There are plenty interesting things you can do with talk pages, but I'm not much of a fan, tbh. We could juts use redirect magic and use forum software on a different subdomain, but that opens up the question of user accounts. So far I think that's actually the biggest issue we'll be having.

The last thing will be bot protection. And look, I am 100% on board with handing everyone who asks a full wikidump. Period. But we also know that the scraperpocalypse is currently in full swing and you'll be flooded by arseholes trying to train their plagiarism machines on your free content the moment you go live. No idea how to handle that. We can't afford a honeypot, IP-banning doesn't work, Anubis is vibe-coded and JS-dependent, so is Gandalf (I also think Gandalf isn't open source) and CloudFlare is a no.

Oh hey looks like the guy who singelhandedly did more for the modern web than all of GAFAM combined has no clue either. Fuck.

I'm still not fully back, but found myself with some bandwidth and idle CPU time. I'll get around to reading all the new articles and participating in the forum soon enough. (At least with my usual verbosity, it's believable that answering takes some time.)

Have we ever explained people how to create new pages? I'm afraid without that newbies might just be confused then.

Probably make a whole new page that is just various instances of these modules for items, tales and other stuff. as they can be configured to have certain tags by default (*intensely stares at ScaledDatura*).

Otherwise, I like it

Anyone got feedback on the new layout? Anything important missing?

I ask Neonach to keep future additions, if any, particularly short. You already contributed more yourself than the rest of us combined.

I created a sandbox for this thing. The first person to stumble upon this thread gets to either
A) Make a name for an Item OR
B) Write the first sentence(s) of the Usage section (the item will be called Placeholder, Y/N, {$Item} or something to that effect then until the article is finished.)

After that, everyone checking by here may append to the article unless they were the last to have edited it. An addition can be anything from one sentence to a short paragraph. (Obviously, keep it coherent to the things others added before!)

Move on to the Report when you feel there's nothing to add to the Usage anymore. Once that one is done, we'll fill out the Infobox in cooperation and start smoothing the edges before this is posted to the mainsite.

(@Meta, @Neonach and @Mario: y'all gonna get replies eventually. Patience is a virtue.)

Also, what animosities? Did I miss something? I thought I was just slowing down for a time in something that started to stress me (which is not what a hobby should do).

EDIT: There seems to have been one message of you I flat out had not read, opened so that it was no longer a new PM, but never read. So, I guess do not leave open letters lying around before going on extended leave? Also, I do not crack ciphers.

Also you underestimate my power, I ate Coruscant Nihilus style before crashing a ship into Mustafar ending the clone wars. So Vader never lost his body, Padme and Sidius were eaten and the galaxy fractured because suddenly all major powers of the clone wars lost their heads at the same time.

Also, sorry for putting a hat on your cosmic horror. I see game pieces, not hopes and story beets.

I also plan to finally finish my Ozor's Uniform Translation, so we have two FR translations; then I'll copy Yorick's ES ones (or rather cross-include them) and add ES and FR to the Projects and Series hub too. (Same goes for Sectors and Steel, probably.)

@Meta: Do you take it personally if Moving Shadows falls under "alternative formats/canons"? If not, I think this might be an agreeable resolution to the recent animosities.

EDIT: Fixed. Doing it with flexbox as is "best practice" didn't work, at least not without restructuring the entire mainpage, so I used absolute positioning which is "not recommended". I think there is something to learn about IT after my birth here.

I believe it's something to do with the way different character sets are being handled. Somehow, somewhere, the individual octets of an UTF-8 encoded string are interpreted as indivudual characters (mind you, this was done after I switched to my Windooze machine, so likely any 125X-charset) somewhere between writing, reading, piping, convertign piping and writing. What happened to the good old days of treating strings as a sequence of byte-sized characters terminated by a null? Oh, yeah, people to stupid to do bounds checking demonized them as "inherently insecure" and act all haughty for generating inferior assembly at exponentially increased compile times…

Anyway, rating modules: so far, the simpelst solution would be just to strip out the rating HTML directly; if it's inside my modified widget, we can also extract the additional info text.

I'm not sure whether can nor whether we should (Hah, take that Jurassic Park!) transfer ratings from WD to MW; assume we get most of our users over to MW, then what? Check who has already voted and prevent them from voting again, unless they are there to check their votes? I think keeping something like a "Legacy Votes" badge is a better solution.

Do we even want to keep the upvote/downvote system the same? I think merely allowing people to "favourite" a page (i.e. give it a thumbs-up, a +1, a star, etc.) is sufficient; if people have problems with it, they should write into the Discussion/Talk page with a call for rewrite/deletion if it's that bad, and otherwise criticism. A sole downvote never helped anyone improve their writing. (Also stops downvote brigading; a phenomenon we never had to deal with but must expect nonetheless.)

When we move to new wiki, let that be the opportunity to overhaul some of our processes.

From what I saw here maybe you need to add

<meta http-equiv="content-type" content="text/html; charset=UTF-8">

Wikidot has the full string as the one suggested by Stackoverflow (sour wine : line 52), while mediawiki does not. Maybe we should just find/replace that in the product.
after f.close():

MetaRep = open(dest, "r+") tempHold = str(MetaRep.read()) tempHold.replace("<meta charset="UTF-8">", "<meta http-equiv="content-type" content="text/html;charset=UTF-8"/>") MetaRep.truncate(0) MetaRep.write(tempHold) MetaRep.close()

…fiiiiiiiiiiiiiiiiiiine. Look at the repo; it's merely 17 lines but it should work quite reasonably. Demonstration — note that much of the uglyness stems from the absence of our CSS, put these side-by-side in your browser's reading mode (F9), and there's no difference.

Stuff I need to do:

• fully filter out all the rating module shennenigans
• Investigate why some, but not all, Unicode characters get corrupted
• Disable auto-ToC
• Test on more complex stuff (tabs, collapsibles, images)

I think this is a promising venue; I tinkered that stuff together in, like, 20 minutes, half of which were related to cross-platform issues caused by MICROS~1, so yeah, that's probably a plan.