Since there have been cases in the past where some fixes/adjustments to the wiki became necessary, and putting that stuff in the discussions or other pages (or worse, chats) isn't worthwhile, here a stickied thread for us.
Eschew elucidation, espouse obfuscation.
From 2022-12-31 10:45 UTC to at least 17:00 UTC the same day, this Wiki, as well as the INTL wiki1 from where we include the current theme had issues with serving the stylesheet. The problem seems to have resolved since then, but problems that go away on their own also come back on their own.
There have been no changes on our end (nor on INTL's end) that could have caused this, and the issue was global (as checked through VPNs and Tor nodes) and cross-browser (I used LibreWolf, Ungoogled Chromium and Tor Browser). The servings failed about 80% of the time, and the periods when it worked were time-correlated (i.e. there were brief windows when it would work everywhere, just to fall back to borked soon after).
I have diff checked the HTML servings from tabs where it worked vs those when it didn't and saw no relevant changes except a reordering of the <script> tags in the <head> element, though no changes in content. I tried resolving all the included URLs from the <style> tags end could load each of them manually.
I have absolutely no idea what caused this, but if I had to guess, it would be some problem with cloudfront (yet Amazon lists no problems in the relevant time frame).
If this returns, we might be able to work around it by simple including an [[moudle CSS]] with a copy of our theme in the nav elements, so it is loaded individually to each page. Not pretty, but it should work.
Eschew elucidation, espouse obfuscation.
Just saw proof that this also happened about 7 hours ago to the RPC sandbox (thanks to EnderDragon101 for posting a screenshot photo of his screen.
So I'll file this one under "yet another reason to leave WikiDot ASAP".
EDIT: Began affecting our sandbox literally minutes after me posting this. The SCP branches are serving their CSS (mostly) through GitHub; I'll look into non-M$ options for us.
Eschew elucidation, espouse obfuscation.
I have to report no issue on my part. The glitch as initially described diapered after less than 10 minutes after I first noticed it and has not reappeared since.
CSS looks fine on all CI-Sites.
Fighting Monsters and Language since 2016.
Yep, appears to be on/off thing. Varying reliability is not good, obviously, but I don't really know what to do about this other than trying a third-party alternative.
Eschew elucidation, espouse obfuscation.
I have disabled the ability to delete pages (for real, the move to deleted: should still work) wiki-wide for non-staff. Rationale:
Authors who still wish to remove their works from the wiki just have to move the page to the deleted: category (Under "+ Options" in the toolbar below every page) and confirm in the page's associated thread that they intend for their work to be removed. We may enact a proper policy on how to handle deletions once there is enough staff around to former a proper consensus.
Plus, you now have to be a wiki member to upload files to categories outside of the default one. Why was that even enabled? At the other hand, non-staff may now rename attached files - everyone already had the rights to upload and delete so I really don't know what the U in CRUD had to be it's own checkbox, WikiDot.
Eschew elucidation, espouse obfuscation.
First, welcome back.
Second:
Is that what is going on? Author self-deletion was not a problem in the past, or was it and we just did not notice?
Yes, pages going missing is an issue, so if you want to start fresh, just tell staff before you pull the trigger. I do like to make backups of forum threads for later reference.
My deletion method worked in the past and I believe with us both here and Baronjoe also being rather active lately we have enough staff to make decisions.
I am OK with the changes in file-policy as I was also not aware that non-members had that right until recently.
Fighting Monsters and Language since 2016.
Alright, as promised in another thread I can't be bothered to find right now, I've taken a look at the feasibility of MediaWiki as wiki software for us.
My verdict: our wiki is small enough and without weird edge cases that we can afford migration. Attempting this with stuff like SCP-EN would be something not even a masochist would enjoy.
Now a breakdown:
Uargh. Setting this up was not fun, mostly because of PHP; the database was a breeze. I used the TinyPHP server for a local instance, and promptly ran into an issue with the International Components for Unicode which… didn't function, except they should have been included. Took my like an hour of fiddling with versions, downloading random binaries, etc. until I decided to screw it an compile them myself, which finally worked.
I refuse to ever be the sole person in charge of hosting the wiki should we decided to get a private server. But honestly, Miraheze (or in a pinch ShoutWiki) probably offers more than we could ever wish for, at less cost and effort.
I've taken a look at the Python Script to translate WD source into MW source and honestly… even I could do better
However, we need to do the whole thing mostly manually anyway, because WD isn't giving anyone API access anymore, meaning we have to manually copy'n'paste the page source and throw it into the script. Then we can also take the opportunity to check the input/output if there are any weird things that require special attention. As we don't have pages that rely on on dark magic (e.g. ListPages, heavy JS for interactivity, CSS animations, other things that really have no place in creative writing) that shouldn't be too much, and leaves the script to be relatively simple (formatting is almost primitive token replacement, most complex is probably parsing and re-assembling tables).
At less than 200 pages, this is something that can be done manually, especially when multiple people help. Get some 6+ hour supercut of some recorded lecture, sigh, and get working.
There are, however, two features of WikiDot which I believe can not be (directly) transferred over to MediaWiki: Collapsibles and and Tab Views, which MW doesn't feature
As these rely on JavaScript, and MediaWiki doesn't require JS to display correctly (huzzah!) they also can't be easily re-created using transclusion or something else. Sure, some extremely contrived CSS gizmo might also get the job done without JS, but at this point, I dare to ask the question: do articles really need these features? Like, these are meant to be documents that can be printed. And the way articles are structured, you're supposed to read all of them anyway, so the argument of "hiding away option content that might not interest everybody" falls away. I consider these an anti-pattern, and even though I used Collapsibles myself in the past (only twice) though, notice that they are mostly absent from our fiction anyway, barring outliers like Elizabeth Kaitlyn or Sporal Contagion which could be rewritten trivially whilst still preserving the author's vision. (Note that they both are rather old articles.)
MW does not have a voting functionality by default, nor does it have a forum (talk pages are such an affront we're just going to ignore them). However, there exist a multitude of extensions both for Voting and Forums which we can use. I haven't tried and compared them yet, but there are plenty to choose from. Jot this down as a TODO. I believe all the extensions for rating use a 3 to 5 star rating scale rather than the primitive Up-/Downvotes of yonder. I have basically no experience with PHP, but can probably adapt the number of stars to 2 and design the widget accordingly. (Just like CLU, I can only alter or delete.) The same thing technically also applies to bread crumb navigation, but we don't use that anyway.
Being made by reasonable people, there is not much opportunity for WD-like in-page styling with CSS. Instead, the whole wiki is styled by a theme. These are default, but can be changed either via URL arguments or for logged-in users in their user settings. (I.e. ?useskin=vector for Wikipedia if you want to revert the horrible change that was pressed against overwhelming contra votes on the world.) There currently exists no theme that suits our needs, but that can be changed; I consider halopedia's Nimbus theme as a good foundation so we don't have to start from scratch.
Eh, honestly forgot where I was going with this post, but, uh… MediaWiki is cool, and I think we should use it. But I'm above getting blinded by the shiny new functional thing and just rush for it like brain dead.
I'd recommend looking more into the extensions and what else is needed to copy our current wiki structure over (and perhaps identify flaws in the current one and try to make it better in the new wiki), and then start with a private wiki limited to the future admins who keep copying stuff over until it is up-to-date with the main wiki, where we will open it as read-only to the public and try to slowly phase it in (and the main wiki out) before opening it for contributions and, upon seeing that it works, beginning to sunset the WD site.
I'm not even trying to sugarcoat this; the whole process is probably going to take until somewhen in 2025. Better than jump-starting and then running out of steam whilst also getting caught up in technological debt and pitiful planning.
Eschew elucidation, espouse obfuscation.
Just noting that CN has contacted me (I believe it was their head admin, but seriously folks, get Discord names that match your WikiDot ones) about the migration and potentially pooling resources.
Apparently they have taken a look at DokuWiki for wiki software and Flarum for a detached forum so far. I consider DW's Markdown-based syntax insufficient, but it is apparently easy to set-up, use and administrate. Flarum is a hard no-go due to being a non-integrated service (compared to MW extensions) and requiring JavaScript to do anything (sure, WD does too, but you get my gist).
Also, it appears that their domain ci-wiki.cn/ is under attack by malicious traffic — so far CloudFlare is keeping them safe (I guess it's just a sweeping assay for vulnerabilities and no targeted attack). Should we consider to self-host (compared to using wikifarms with their own anti-spam/Denial-of-Service measures) keeping that stuff in check should also be a thing to keep in mind. I'm by far no fan of CloudFlare, but they do deliver results. (Also, claiming live and on stage that "You underestimate how much money we have in the bank" gets them bonus badass points.)
I'll continue testing with my own local install — I believe I just broke its database due to being somewhat … zealous when clearing up "unused" memory. Another reason you shouldn't have me adminstrate anything on my own, else we'll end up like yugipedia.
Eschew elucidation, espouse obfuscation.
I see you are hopelessly incompetent at your job, because else you would have realized that there exists a very simple Extension which does Tabs and Collapsibles 2-in-1.
It requires JS though, but unlike WD's approach, is actually fairly useable without (as in: I've used a wiki with it for quite some time before I realized it was technically broken due to uBlock origin) and thusly preferable.
Eschew elucidation, espouse obfuscation.
I am reaching you through a prepaid mobile hotspot provided by everyone's ISP of choice — Tchibo. These circumstances which force me to evaluate the necessity of every single byte transmit and way to much free time made me consider the current role that collapsible blocks and tab views have on our wiki.
Namely, that the main, fiction part of the wiki could probably do without them.
First of, let's consider their intended use: make a document easier to read and visually appealing by hiding extraneous information — only show the radiation pattern of the antenne I use, only show the economic details of the province I care for, etc. and neatly hide the rest.
I'd argue fiction does not need this — if information is superfluous, why are you putting it into the article in the first place? There are no "situational" passages here, simply bloat.
Besides this honestly abstract point, there are a couple of more practical ones:
I have looked out our Arsenal so far (are there any tales with either?) and found that we could universally replace them with minimal changes. See for example the "individual sheet of paper"-styling I've been experimenting with. Give that a snazzy heading and perhaps a "water mark" (easily doable with CSS) and you acheive the same narrative effect as collapsibles are currently being used for, perhaps even more appealing.
So, to ask the question again: can we bring the wiki in a state in which people could easily print articles without having to worry about collapsible content?
Eschew elucidation, espouse obfuscation.
As long as I can keep my sandboxes tabbed this should not be a problem. We can try the no tabs approach, but isn't that "sheet of paper" just a collapsible with a different default state, as such still posing the problems mentioned above?
Found this:
[[div style="background-color: #F7F7F7; margin:1em; border: 2px solid #5C5C5C; padding:1em;"]]
[div]
Might be usefull.
Collapsible and tabs are also used to differentiate between different clearance levels.
CI-Cases that use Collapsible and Tabs (more then others):
Sporal Contagion
Elisabeth Kaitlyn
Creative Cat
Shanalick
Sekhmet
Analiana Matos
He's Got The Whole World In His Hands
We Are Slaves Of Creatures We Cant See
Eternal Selach
Sadistic Statistics: The Infobox's Guide
Fighting Monsters and Language since 2016.
Zoinks! Foiled my plans once again!
(Yeah, I forgot I changed 'em to collapsibles. Lots of my old stuff contains sins I try to get away from, like using footnotes and those dumb login-interface thingies.) I see you already found a way to just use it for divs. (Yet I believe the collapsibles with a default display are actually functional without JS, except that you can't close them.)
I'll take a look at getting a working watermarked div running for the fly boy as a proof-of-concept. Probably doable today or tomorow, but right now I can even less stand to my word than usually (still relying on tchibo where Telekom failed me).
Sporal Contagion in particular was what made me think about this in the first place.
Everything that isn't a published article, be it something in the sandbox, a forum post or an administrative page is fair game imho. In the end, a properly-indexed page with table of contents might be the way to go for most of those (except my rather extensive replies, of course), but those probably aren't what people are coming for here.
Eschew elucidation, espouse obfuscation.
Not sure which plans you mean.
I actually agree with you so what did I foil? The method?
Also, I hope that you can get some better connection soon. The tchibo way might be a radical new approach but it can not be sustainable. I wanted to make a "Get better (internet) soon" joke here, but it came out more demanding than a "get well/better soon". So now you get an explanation to a joke that did not work (as intended), instead of the joke itself.
Should I edit the older stuff with the div blocks? Should this only be a "from now on" thing?
Fighting Monsters and Language since 2016.
For the protocol, I'm using a mobile hotspot acquired at Tchibo, not just sitting in a store leeching free Wi-Fi. (That thing has a SIM-card but didn't require ID. Now I'm curiously looking at E.164 mapping, wondering if I just found a loophole…)
Anyhow, if you don't want to wait for me, you can just go ahead, though I wouldn't run the whole wiki through the process before we got some basic acceptance testing done.
But if we were to do this, everything had to change, else there is no point in doing it.
Eschew elucidation, espouse obfuscation.
I went ahead and deployed the "watermark" replacement on Sky Nomad. Won't work on longer text passages (unless you manually dupe it) and looks horribly on mobile, but for me, ona desktop browser, it's good enough. Thoughts?
Eschew elucidation, espouse obfuscation.
A little faint. Can you make it "faded stamp red" (#d75149)?
My firefox on mobile does not like wikidot on principal, but there it looked at least serviceable. It splits "Alpha Eyes Only" across 3 lines, was that what you meant?
Fighting Monsters and Language since 2016.
Well, watermarks are by definition somewhat faint, and I'm massively aware that making it to strong can severely impact readability, especially for people with seeing impairments.
I believe you are using a Samsung product then. Checking with Ctrl+Shift+M only those devices where generating three lines, all other had an ugly two (mayhaps I should use text-align: center; for both cases).
I asked the CSS gurus of SCP-INT if they have any ideas, so let's wait and see. The easiest way to make this thing tiling would be just using a transparent background image set to repeat vertically. But people who use images of text instead of text deserve to be court martialed, so…
I might consider adapting this for the CoR as well to check it for a longer passage of text.
Eschew elucidation, espouse obfuscation.
I have moved the City of Ruin to MediaWiki to try how difficult it is to replicate all of our Wiki's feature's (minus the collapsible rating module/"Anubis") and how difficult it is.
For something like the Unisnger, it's actually a mostly straightforward process (but we definitiely need a script for tables; doing this manually is maddening!) with the correct layout of images for both desktop and mobile being the main problem.
However, I am currently working on Elisabeth Kaitlyn, who relies much more on additional styling, which requires significantly more manual adjustment than my find+replace approach). Still, I believe to have shown once and for all that migration, at least in a read-only state, is totally possible.
Also check out Unsi in Minerva and Timeless (note that the former automatically un-collapses collapsibles).
Eschew elucidation, espouse obfuscation.
Is there a way to hide the "edit source" buttons?
The white is also to white. I like the CIs #f3f3f3.
Nice work anyway.
Fighting Monsters and Language since 2016.
Elisabeth's done. Note that unlike with her WikiDot version, she remains readable even when you disable JS in your browser — meaning we can have our cake and eat it too.
Well, hiding them with CSS is no problem, I'm afraid the massive spam of them in the CoR stems from me having used Headings for formatting, which tell MW that this is a new section. I'll go and ask their support, mayhaps we can do something about that. (Also note that WikiDot has an "Edit Section" thingy under +Options; it's just nobody used that.)
For now, the themes are just random default ones I tend to swap in whilst we have none. Take a look around Miraheze to find other wikis with designs we can base ours off of. Besdies the Gay Ravens there is the Land of Liberos Project which uses agreeable colouring. There is also the Cadence which is apprciate for their techy, yet clear design, the When They Cry Wiki which is well-laid out. Formerly, I've been already suggesting Halopedia, but honestly, outside of their main page, it's kind of attrocious?
Mind you, all these themes can, if properly implemented, be override through either a logged-in user's settings, or through an URL agrument (?useskin=) meaning everybody can run what they prefer; like a dark mode.
We could start with a stole theme and work from there, or we simply hire somebody to do it. I'm not opposed to spilling a few bucks to be some fledgling web designer's first portfolio entry, if we get a professional, maintainable and accessible theme in exchange.
A terribly hacky python script to transfer most things to MW syntax is probably possible, but will require manual post-processing on more extravagant articles (like EK); but tales and simpler Items should be fine.
However, said script would require a large chunk of WD-syntaxed Page sources to churn through… since WD offers no API access even to paying accounts anymore (is that even legal?) this means: time to web scrape. I'll probably run Selenium for that, but perhaps you too want to make a full backup of all the wiki's sources?
Eschew elucidation, espouse obfuscation.
From the wikis you suggested I would use "When they Cry". I am still no CSS wizard, but changing the background to a more CI picture and replacing #7e5b07 with #f72b34 and
#deb887 with #e72f37 should get us pretty close to our current theme.
I still think that a simple find and replace could get us there, but I do not know the miraheze format, so I do not know how "ci-wiki scrape html" translates to it.
I will make a more recent backup once I stop getting DB200 errors every ~20 read-attempts. Usually they are fixed after a reload, but I do not think it would bother the scraper so a few pages might just get lost.
Fighting Monsters and Language since 2016.
Perhaps parsing straight from HTML might even be the more "elegant" way. WD has loads of overloaded Syntax (i.e. the double dashes -- which can apepar as an individual character, beginning and end of strike-through text, or as part of a horizontal rule.
Ideas for a "more CI picture"?
I'm actually hyped for MW's template magic. I already found plenty of useful ones, like Template:Dialogue and it should be possible to do stuff like the Infoboxes and a couple of other frequently used formats (video transcripts, PoI Sheets, etc.) templated too. Not to mention simple in-line ones for stuff like showing the Rank symbols from the Personnel Designations page. Or Dynamically creating a set length of redacted text (see the failed, JS-dependent experiment in the Kernel).
Just, like, don't make me write an extension in PHP. I know I claimed I could easily work myself into the language, but that was before I learned that PHP is a programming language made by people who can't code for people who don't want too. Seriously, the fact that in PHP 0x0+ 2 is 2, but 0x0 +2 or 0x0+2 equal 4 should make you drop everything an run away very fast immediatly. Not just do I revoke my previous offer, I will also deny under oath to have ever made it.
Also, the Vault's lilac-apricot theme will be inevitable.
Eschew elucidation, espouse obfuscation.
My first thought of CI-Picture was the site default gray, than I wanted static, black and white hexagons. In the end I propose this, with maybe a grayscale and a noise layer. We could also make it something CI original, I would have to go looking deeper or create something using my art. Also, this rotated 90° would be funny for at least 2 minutes.
Fighting Monsters and Language since 2016.
Oh, I certainly have my own collection of stuff as well. I appreciate your slightly rustic "weird science" take, but perhaps we might also go for something more totalitarian? I have a great unsorted collection of links from whenever I went on vaulting sprees…
And I remember uploading the latter image. It turns sinister suprisingly easy once you edit out the 5 in the maximum occupancy sign. What about the photographer? Why are there more chairs than 3? Y'know what? Why do you get to have all the fun with the dumb challanges? I'll try to cook something up until my birthday, that's it.
Eschew elucidation, espouse obfuscation.
Its nice to see progress even though I don’t understand where exactly we are right now. How about some more abstract ones? Basically, the VDD or the ones found in Item field. The totalitarian one could link to the Facility list.
I went ahead and changed the top bar to have a new "User Tools" drop-down. This is where the button to apply the old license can be found in the future. It also temporarily has buttons to add the Migration Challenge Levels for (mostly my own) convenience. I'll remove them once it's done.
These challenge levels are (hidden) tags I'll use to determine how easily a given page's source is parseable to wikitext. We'll ignore rating modules for now.
For stuff (mostly tales) that only employ the most basic markup: bold, italics, headings, horizontal rules, [[[Internal links]]]]] or [*https://example.com hyperlinks]]]. (No underlining though.)
Stuff that includes everything from CL0 plus tables, blockquotes, footnotes and normal images.
CL1 and collapsibles, tabviews and coloured text. Also probably most stuff with basic [[div]]-based styling.
Everything that uses weird stuff that is easier transferred manually than considered in the python script
To indicate we saw a page even though we don't actually need to assign a CL.
I'll go ahead and apply some of these already, everybody else is invited to help according tot he rules above.
EDIT: Ugh, PoI files are gonna be fun…
Eschew elucidation, espouse obfuscation.
CSS has support for "Custom Properties", i.e. variables. (Which have an absolutely horrid syntax — what is it with web standards and the fetishization of double dashes?)
It'd be advisable to set a couple of "colour groups" or whatever which our theme (and future themese, both an WD and MW) could simply alter, to support, for example, dark modes. Somethign like --insu-bg-colour and --insu-text-colour is obvious, but I wonder how much we need, and with how many we can get away from. E.g. we need a special colour for our quote blocks, but is it the same used for many buttons? There is definitely going to be a colour for links/interactables, but is it the same colour as we use for eye-catchers (h1) and accents (like in the rating module)?
Just something simple to mull over whilst I procrastinate.
Eschew elucidation, espouse obfuscation.
It appears that Miraheze is undergoing some internal re-structuring. As far as I can tell, they are currently incorporated as a Ltd. in the UK, which is kinda dumb if you are acting as a non-profit. Anyhow, it appears to be gravitating towards re-establishment as a true non-profit Foundation (hold the quibs).
Not really a sign of worry, except … this would entail a move to the United States. What would that mean in regards to migration perspectives? I mean, the infrastructure would probably stay the same, but the jurisdiction changes. I don't believe there is any concern regarding our user's privacy (The UK is a 5EYES nation anyway, and unless you're really dumb your Wiki account won't contain anything besides a disposable addy and your public interaction on the public wiki.)
However, whilst the british web legislation has been continually slowly getting worse, moving to the US is much more explosive: depending on the individual states and depending on whether the US population will vote for an orange man child, we have the very real potential of suddenly ending up in a legal hellscape with laws that we simply could not comply with even if we would. So it's up to chance whether we'll end with better-than-british legislation or having to bow to some carrier who wants to track our periods.
Sadly, when it comes to free wikifarms, I couldn't find any on EU soil. A paid provider would still be possible, as is self-hosting, but I thought I better bring this up here and now.
Eschew elucidation, espouse obfuscation.
…and they merged with WikiTide (in the US). I'm not opposed to that if it weren't for greedy messages begging for donations immediatly appearing on some wikis. No idea if this can be disabled on a per-wiki basis, but it doesn't appear that, hiding that stuff through other means is against TOS.
And, y'know, I'd actually enjoy having a platform I can financially support instead of an economic zombie like WikiDot that relies on long outdated ad platforms.
Eschew elucidation, espouse obfuscation.
Zombies might not be as stable but at least they will not suddenly change their TOS.
What ever keeps Wikidot alive is most definitely known to the person running it and considering the size limitations and the hardware I estimate necessary to run all of Wikidot I would say that even if just 10 people payed 5 euro a month the zombie could keep going for a while.
Wikis without financial backings are limited to 300mb and 1TB hard-drive sits at 40 euros new.
The large beasts like the foundation have paying membership everything else might just as well be pocket change.
Wikidot is probably 1, maybe 2 old servers that are probably not even rack-mounted. It is basically as close as you can get to passive income. A domain, internet and power is pretty much all you need and a server that is trimmed for storage and just basic applications might just run on less power than it would take to run an incandescence light bulb.
Fighting Monsters and Language since 2016.
Except that this Zombie is doing what rotting corpses do best: slowly falling apart. Ignoring the two months we paniced over a PRETTY SEVERE exploit, the nigh-absent documentation and completely outdated and customly-tinkered framework, there is still the fact that essential things do not work:
We have no search whatsoever (whenever I need to search something, I grep my local backup), no API access even for paying customers, and no customer support — last time there was a problem the only guy in charge of WikiDot had to be contacted over a microblogging service that's incredibly unstable itself.
As I mentioned, I could trivially bankroll the Wiki myself without prior notice; the problem is that I am in no position to actually host stuff myself. My ISP only gurantees me 97% uptime without even the chance of upgrading to a business plan, and there is no viable way to anonymously buy a domain name for me.
Plus, somebody would have to do all the boring behind-the-scenes managment (setting up the database on which the whole shebang runs, maintain the actual (sub)domain, get us fail2ban or something to prevent the background noise of script kiddie attacks, etc. And from my vantage point, you don't seem terribly suited in that regard (who just wanted to run a server with a Hard Drive?). Since Cad never shows up anyway, it would all just end up being my problem anyway. And I'm not planning to end up burned out after three weeks doing something I'm only moderately qualified to do.
You again make unfounded claims about computer science. A processor is basically equivalent to an electric heater of the same wattage. So it is an incandescent lightbulb, just not driven to the degree of incandescence (which is mostly a function of thermal mass here). A two-socket x86 servers with up to eight core chips will run at rest somewhere between 150 to 200 watts. At full utilization, this energy usage can double to as much as 400 watts. Clearly, we are discounting cooling here (as this setup would probably operate in somebody's basement, not a big server farm).
If you have a better alternative to Miraheze, come forwards with it. ShoutWiki is still a thing. But simply in order to not keep all our eggs in one worn and battered basket.
Eschew elucidation, espouse obfuscation.
You do not want to do customer support and they do not want to do customer support. I would probably try and be bad customer support.
You can run servers from hard drives just fine. It just depends on what load times you are willing to stomach.
I know people run servers of Raspberry Pis. A wiki that is not payed for is 300mb (like the Insurgency, 900 if we take the sandbox and image database into account). With our user numbers we might just run the whole thing on a 2GB Pi. I actually never tested how far these little buggers can go. I correct my stance. A server as you propose would be 5 100 Watt bulbs not one.
I am not currently in a position to just build a DMZ into my current network and just set up a Moodle on a Pi. Moodle was just the platform I worked on during my education and the one I know somewhat how to set it up.
As you said, in the end it is the problem of whom ever takes up the migration and I just can not be that person at the moment.
Fighting Monsters and Language since 2016.
It's not just that I don't want to, I'd also be at least as bad as you. You know how I am when I explain things to other people.
Hold back the "servers with HDDs" thought for a moment.
I think your estimation of wiki size is fundamentally flawed: WD only calculates attached files. Stuff like page source, it's complete revision history and the forum do not count towards the limit. And if you take the main page, at ~4k characters and ~300 revisions, that's already about 1.2 MB for a single page. However, most of our available space isn't (meaningfully) used anyway.1 My current Backup of this wiki is roughly 300 MB, and this includes an abhorrent amount of different wiki thumbnails (I was to lazy to exclude those) and duplicates of all the necessary (and unnecessary) JS for many pages.
Now, the interesting thing about that is that we could probably run the entire thing as a RAM server, using persistent storage only for backups.
However, having an own server and a domain would allow us to do some fancy other things. Like hosting our own matrix space, our own booru (you wanted that, if my memory doesn't fail me), our own git or even stuff like an own Invidious instance (which costs money, but then again, if the artemislena girls can afford it…)
Does anybody outside of Germany actually use Moodle?
I'm also in no position to host a stable server. And this also includes physical stability — if some of you could see some of my setup, you'd be accutely aware I live in an earthquake zone. (Thank goodness my dog is afraid of walking stairs too…)
Eschew elucidation, espouse obfuscation.
Through, oddly enough, the old Glitch City Laboratories, I discovered SMF, which apparently integrates well with MediaWiki.
@Meta: I checked the options, and it appears that you can form a "nicht-eingetragenen Verein" which seem to be mostly what we need — minimum of 2 members, no need for in-person meetings or any fancy roles and reporting, merely an entry at the Finanzamt (which should be private, afaik). Liability wise, that entity is not rechtsfähig, though, i.e. all contracts/other stuff must be through one of us members. (I wonder how this interacts with the Impressumspflicht.)
Eschew elucidation, espouse obfuscation.
I do not like the "every member is fully liable with their private fortune". If we do this we need to limit liability, especially if we want to continue running a "wild west sandbox". I read that it is possible in the "Vereinssatzung". We should also go for "gemeinnützig" if we can.
If we can subtract CI expenses from taxes we might not even be paying extra.
Fighting Monsters and Language since 2016.
Liability is only for members, i.e. people paying membership fee i.e. most likely the two of us. Whilst I still would prefer to be able to machinate from the shadows with no hostile means of recourse, we should be relatively safe - afterall, it is possible for private individuals to run Tor exit nodes, Invidious Instances, etc. without being themselves on the hook for the plethora of illegal content there.
As long as we keep an eye on the sandbox (I do) and remove violating content (which we would do anyway, because of WikiDot's TOS), it's realistically not that different from how we're acting right now, and given that we'd operate from Germany, we'd probably have higher standards towards privacy and fReEdOm Of SpEeCh than from WD.
Gemeinnützigkeit is as far as I can tell a big hurdle with tight criteria to meet - a hobbyist association certainly won't meet it without offering plenty of other services (I'm thinking of Git hosting, privacy frontends, Tor relays, etc. which require attention and money) and I believe that by the time it's the fiscally responsible thing to do, we can and should already become an e.V. instead.
Eschew elucidation, espouse obfuscation.
I would not count on save-harbor-laws. Politicians try to kill those frequently and I do not want to suddenly be on the hook because a law changed while we were not looking.
Doesn't your sandbox with the vault boldly claim to use images incompatible with our license?
We pretty much are an art collective so we could qualify under §52 ab5. With the money going through this construct being:
The additional accounting might not be so bad. We can whittle it down further by renting a VM, but I think we can get our finances in order even with "Gemeinnütigkeit".
Fighting Monsters and Language since 2016.
Your legal analysis is wrong to the point of gross negligence.
First of all, safe harbour laws aren't what you think they are. They mostly concern data privacy and where you can legally store and process data. Y'know, the whole reason Amazon is currently dumping billions into an AWS European Sovereignty Cloud which is 100 % located and staffed in actual europe. Whilst there are in fact attempts at throwing over these safe harbours (in particular the monarchists, if I remember correctly), that is no concern for us as our privacy policy will be along the lines "we probably log IPs for fail2ban and else just stuff you intentionally post there".
What you probably mean are adoptions of the WIPO Copyright Treaty, in particular the american Digital Millenium Copyright Act and the european Copyright and Information Society Directive 2001, which rule that as long as you merely "curate" user-generated content, and abide to takedown requests, you're as liable for that stuff as the owner of a kiosk is for the content of the magazines they sell. These laws are under mediocre bombardment from various rights (hostage-)holders, but have an even stronger counter lobby in the form of big tech, whose business model relies on those protections. I don't think that "politicians try to kill those frequently", at least not compared to E2E-encryption, but even if they were, believe me, we would know in advance because it'll be a massive shitstorm.
Next up: I "boldly" claim to be hotlinking images without intent to publish them. You really think I'd be brain dead enough to knowingly and publicly do something obviously illegal? Mayhaps you are projecting… Anyway, hotlinking, whilst rude and usually frowned upon due to the bandwidth theft (I believe that 99% of the traffic to my sandbox is just me), it is perfectly legal, assuming the content you're hotlinking is, in itself, legal. After all, it only tells your browser to go to that URL to fetch that resource, and doesn't make an actual copy that could infringe on copyright. This is a given in the EU thanks to a binding ruling of the European Court of Justice from 2014 — however, I don't blame you for getting it wrong, as a really dumb reversal from 2011 decided that hotlinking is infringement, even though it was ruled okay since 2003…. Anyhow, the EUGH ruling has been adapted here too.
So why did I go out of my way to explicitly mention I'm not publishing them? Because licensing stuff you don't actually own is kind of illegal? There is a massive difference between saying "hey, this is an image I used for inspiration" and "this is an image of my Item, which therefore is part of the article and thusly given to you under a CC-BY-SA license". (Also, I just checked, and surprisingly many of those are actually from WM Commons, huh.)
I don't think that taking care of our own physical server through a most likely non-commercial IP connections ("97% uptime guarantee" trololololol) is wise, nor cheap. Using literally the first search result, Ionos rents you a server for 7 USD a month — and as a wiki is basically just a database serving mostly static webpages, even their smallest option is going to be dreadfully bored (= opportunity to run other stuff).
Renting a server has multiple advantages:
Since we'd be simply using some FLOSS wiki software, there's not really a risk of vendor lock-in. Worst case scenario, we dump the database and go somewhere else. I believe that all other perceived disadvantages are of the "raving paranoiac detached from reality" flavour if we stick to a privacy friendly hosting provider. (Reminder: we are currently putting out faith into WikiDot.)
I'm the swabian here and even I don't get why you're so insistent on getting some foreign ccTLD instead of going for something reasonable like .org, .net or perhaps even .wiki (but please no cypriot domain hacks). If you really think a couple of euros are really worth using some non-standard TLD (which is probably going to cause reputation issues and/or make us seem illegitimate to users), I once again offer you to fully pay the difference out of my own pockets. Like, what if we one day get a dutch translation wiki?
Eschew elucidation, espouse obfuscation.
You seem to have things well under control. As long as we limit my liability I am good (and not a lawyer).
My main concern is my money and stable operations.
Fighting Monsters and Language since 2016.
(Not sure where my last reply went, but oh my. Here's the quick version.)
We're very certainly covered by §7 DDG, but I believe we still need to copy a suitable preprint disclaimer somewhere (think the standard privacy policies to comply with GDPR).
However, I'm afraid the n.V. can't be in the Imprint, due to having no juristic personhood. Alternatives would be checking in other EU countries, perhaps? Risky.
Eschew elucidation, espouse obfuscation.
I actually got it from the site you linked that started my liability tangent (check for "Haftung"). There are two things I would read from this:
My comments about Gemeinnützigkeit also came from the article:
strg+f "Kann ein nicht eingetragener Verein gemeinnützig sein?"
It reads as if we just need to modify the Satzung a bit (I assume we get a proven Muster and just adjust it and then sent it to a Notar to get us on the right side of the law, before it goes as a mail to the responsible Finanzamt, which sends us back our certificate and than we need to send something every 3 years to keep it). If I read it correctly it would get us out of taxes (Freigrenze).
Fighting Monsters and Language since 2016.
First, taxes: As far as I can tell that one works basically like a private persons Steuerfreibetrag: if we can assure that the whole volume of cash flow is below a given value, it's free. I'm not sure thought if this only income, or overall money movements thought.
For "wirtschaftlicher Geschäftsbetrieb", I found this page:
Unter dem wirtschaftlichen Geschäftsbetrieb einer gemeinnützigen Organisation versteht man jede selbstständige nachhaltige Tätigkeit, durch die Einnahmen oder andere wirtschaftliche Vorteile erzielt werden, die kein Zweckbetrieb sind und die über den Rahmen einer Vermögensverwaltung hinausgeht.
I don't think we'll have any income (besides "membership fees", which probably don't count?) until we actually start selling stuff (probably merch, since I think neither of us wants banner ads). If you check the (shockingly long) list of examples there, I don't think any applies to us.
Notaries are insultingly expensive and as far as I can tell not required? We can do that, but in this case it'll be your money (the same way I'll be paying for a sensible domain).
Usefull Links:
Eschew elucidation, espouse obfuscation.
It appears the Tag buttons also work on non-wiki members, as long as they are logged into WikiDot.
For now this is not a problem since it has been only triggered by one person, most likely on accident, but yet another reason why we can't leave this godawful shithole soon enough for something that actually works.
Yes, I don't even bother the use euphemisms anymore.
Eschew elucidation, espouse obfuscation.
