Cool, yes please do!
Or contribute to the development of Interstellar, Blorp, Voyager, Thunder, etc. Honestly most people here seem to use 3rd-party apps rather than a web browser UI, which I’ve never really understood exactly why but I’m just saying that they do.
Used to be worse yeah, although again, it might be better now? And it was fixable by just rebooting every so often.
Overall people say both that the code written by the PieFed devs is much higher quality… while others say the exact opposite, so I have no idea:-P.
I hope you are happy with Mbin. At the very least it’s nice to have options to choose from. I don’t really like Mbin personally - nothing against the project but the web UI is much better suited to Mastodon-style content (like with long titles that wrap around the screen so don’t waste that horizontal space) than to like cat pics with super-short, maybe even one-word titles. And it has odd terminology and procedures like upvotes exist but they don’t affect sorting like they do in Lemmy and that is rather done by “increases”, which I guess are somewhat like Mastodon “boosts”? (I’ve never used Mastodon so I have no idea) Also, those are publicly visible whereas the downvote equivalent “reduces” are hidden from view. So… it will take quite some getting used to!!
You are quite welcome!
You can use a combination of stability and age - Uptime and Months, columns #2 & 7 in the linked webpage. High uptime % in a well-established, long-running instance is the best.
One thing to note is that the implementation language matters significantly less than the method of handling database accesses. According to the sh.itjust.works admin team who announced an upcoming migration from Lemmy to PieFed this year, Lemmy’s handing (which used to be fairly crappy, requiring routine machine reboots rather than solving the issue of leakages) has recently gotten much better.
Another reason PieFed performs so well is that it is 25-fold more efficient in pulling data from its servers than Lemmy. That is a design choice that hopefully one day Lemmy will catch up to making also.
Thus, the choice of back-end language - Rust vs. Python - is much less relevant than all of the other factors involved. Although it does result in very noticeable downstream effects, particularly the increased speed of adding new features.
I would even go so far as to say that the choice of software itself - PieFed vs. Lemmy, or Mbin, nodeBB, Mastodon integration, even mostly-dead (though reportedly undergoing revival!!) projects like Sublinks - works out to offer far less of an impact to your overall experience than the choice of admin team. e.g. the .world and .zip teams both offer both Lemmy and PieFed options, and if you use something like Voyager (which only offers “experimental” support for PieFed), then you will barely notice the difference.
Your other questions are answered already but I hope this additional information has been of interest as well.
Edit: another consideration is whether you want to host your own instance. For that I would unhesitatingly recommend PieFed as by far the easiest to deploy (especially via the docker container) and using least resources. Tbf, that is not from 1st-hand experience but so very many reports of others do state this.
Fair, that is the part that I know the least about but at a glance… it does look like you are correct. I will edit my comment above to reflect.
Yes, I am glad for db0’s correction. I already edited my comment that you replied to here but our messages got crossed while you were typing, anyway thank you for this clarification as well.
It seems like instances on ptz’s personal shitlist need to use db0’s fork from now on, or like blahaj just abandon Tesseract altogether (as you pointed out, the author even asked for thus:-P).
Thank you, this is an extremely helpful description.
Btw neither fedinsfw.app (with 2.4x higher number of MAUs than PieFed.social, making it by far the #1 most-used PieFed instance), nor PieFed.zip (with self-reportedly 541 MAUs, making it the clear #3 PieFed instance) show up in that instance picker list.
I tried viewing the list from multiple instances including from PieFed.social itself. So I’m not sure how much people actually make use of that picker, and it seems buggy as hell (unless both of those instances have intentionally decided not to be listed?). Though even more confusing is how PieFed.zip also does not report onto https://piefed.fediverse.observer/list (though Lemmy.zip does appear at https://lemmy.fediverse.observer/list).
Anyway, it’s good to know that while PieFed does not have a hard-coded instance blacklist - nor are those 4chan filters hard-coded either iirc - it does have an extremely odd and unhelpful depiction of what constitutes a “good” level of defederation on that instance picker list.
And now with the highly controversial and extremely bad roll-out of the vote quota issue, while I would not go so far as to recommend that people steer clear of PieFed itself (instances like PieFed.zip and quokka and blahaj seem to be doing just fine with it, e.g. maintaining a watchful eye on the opinionated additions to the sourcecode), I would unequivocally recommend avoiding trusting the instance picker at this point. As you showed and I piled onto here, it is just an absolute mess (and Rimu is showing strong signs of burnout so unlikely to be fixed anytime soon).
Thank you for the clarification and for making the fork, that seems like it will be very useful if you want to continue with that front-end UI.
Edit: I edited my statement above and will try to do similarly for others where I have been wrong.
The first list is not hard-coded, as it depends upon an instance-set environmental variable. (Edit: I retract this statement as I may have gotten confused between the differences between the two lists.)
This other, hidden list though… that one is a lot more dicey. It can also be disabled wholesale, but is not easily discoverable nor editable, it is essentially his personal blocklist shared with everyone who uses Tesseract. (It could be nice to open that up as a more fully fleshed out feature where instance admins can more readily control whose personal blocklist gets pointed at, or even allow an individual end-user to input one?)
Though his latest manifesto says that it can be disabled, both by instance admins and by individual users (Edit 2: or maybe not? Admins can disable the ability to users to UNBLOCK the content, but not as readily simply disable the blocking in the first place?).
Anyway since it can be disabled, it’s not a “hard-coded block list”, as in mandatory.
Precisely. We could point people to the very dictionary definition of the word, but it seems not to change many people’s mind who have already decided which “camp” they want to be in, and that they will (attempt to) preempt the word to mean… whatever they feel like.
Hard coding is the software development practice of embedding data directly into the source code of a program or other executable object, as opposed to obtaining the data from external sources or generating it at runtime.
Hard-coded data typically can be modified only by editing the source code and recompiling the executable
Therefore, a statement like
IF (instance-set-variable) THEN
{
... implement block
}
is by definition not “hard-coding”, bc it asks the instance admin what value it wants for that variable, and then only conditionally runs the code to set the instance block if the admin asks it to.
There are plenty of things to criticize PieFed for, but “hard-coding instance blocks” is not one of them. (And now it seems the same error was made regarding Tesseract. Edit: this point is terribly unclear but probably not correct, as in one of the two blocklists is definitely not hard-coded but the other seems to be.)
There is a lot of misinformation about this issue.
Tbf, Tesseract reporting the message that it is “incompatible with that instance" seems valid criticism, that should definitely have been handled MUCH better.
Whether the OOP might have been right or wrong remains highly unclear when most (edit: some) of their claims have ended up disproven, which just makes finding facts that much more difficult. That said, I do sympathize narrowly with the frustration they must have felt to have to have discovered the issue the way they did, with no announcement having been made.
Thank you for offering greater precision here in your own responses.
I am not saying that you should switch to use or even like PieFed, but afaik there is no feature that is “trying to trick users into thinking there’s a technical issue with servers”, or at least not one that Lemmy and Mbin do not identically share. I would be happy to read about one though if someone wants to point me to it?
e.g. just a few hours ago I recall reading a report where someone was confused why a post did not show up for them on PieFed but it did on Lemmy. The reason turned out to be that the user has been banned on PieFed.social. The key here is that the latter is an instance (a server machine) whereas PieFed is software.
Lemmy instances also ban people too btw. Some Lemmy instances are in fact quite (in)famous for this.
But that is not “trickery”, that is simply PieFed (or Lemmy, or Mbin, whatever) working as designed, someone’s lack of understanding notwithstanding, and likewise for people who want to cast aspersions onto their tribal “opponent”.
All that said, PieFed does have a weird way of blocking all users on an instance. If the devs of Lemmy, PieFed, and Mbin could get together an agree on a protocol whereby a signal could be sent that all software platforms would understand and know how to handle… then such maneuvers would not be necessary. Instead, legacy software like Lemmy are not always going to be able to keep up with PieFed’s rapid pace of development - e.g. PieFed offers numerous things that Lemmy simply has to shadow-block (with no report of having done so), including polls, user and post flairs, various special features of communities (e.g. automatic deletion of submitted images after a certain time period, usually one year), etc. To be clear this is not “nefarious” activity on behalf of the Lemmy devs, it’s just that the software doesn’t know how to handle such content, nor do the devs make it a priority to make it do so (why should they?)
And likewise, the ability of Lemmy uses to reply to PieFed users that have instance “blocked” them is one such source of friction between the two software offerings.
If you are thinking of creating a community, it matters far more what instance admins you choose to interact with than what software that instance is running. Especially as some (like quokka, and reportedly soon-ish sh.itjust.works) switch around, while others (like .zip and .world) offer literally both options (Lemmy & PieFed).
I seem to be OOTL, what is the situation wrt blahaj?
I have been known to be a little hyperbolic myself - it does get in the way of clarity, unfortunately. The social media style of communication lends itself more readily towards “slamming” one’s “opponent”, as opposed to calm rational dialogue to arrive at a truth.
No, Tesseract is a Lemmy alternative front-end UI, it has nothing to do with the OCR, besides sharing the same name.
Also, Tesseract does not actually use a hard-coded list - that turned out to be misinformation by the LLM-written OOP. In actuality, there is a “toxicity filter” that admins can toggle on or off as they please - the polar opposite of a hard-coded feature. (Edit: there seems to be two lists, and while at least one is definitely not hard-coded, the status of the other is less clear, at least to me, and might be hard-coded.)
Don’t believe everything that you read on the Internet - it is not all true.
That said, Tesseract is quite opinionated in its views, and instance admins who hold different ideologies than its author might want to actually read about which settings they want to have toggled or not, rather than merely accept the defaults.
It does not help that for that issue, as too for this OP, a lot of misinformation is spread, which obfuscates the truth.
e.g. in both situations, people claimed that something was hard-coded when in fact it was not. (Edit: this situation is terribly unclear, and one of Tesseract’s two main sources of blocking actually is reported to be hard-coded, even though another source turned out not to be hard-coded.)
That doesn’t mean that PieFed isn’t actively censoring content (it is, and misleadingly so too), or that you should use it (use whatever software you want), but it does mean that the people criticizing it, having cited that specific “alternative fact”, were spreading lies. Probably without realizing it, yet they certainly don’t seem all that concerned about actually reading sources in order to find the truth. Which, yeah, complicates people’s decisions about what services to use.
Downvoted for helpfully answering a question, only on Reddit Lemmy.
Edit: this situation is terribly muddy but it seems that while one of the blocklists (the hidden, larger one) is not hard-coded, the other (smaller, instance-only?) actually seems to be.
Actually… it is toggleable, though not customizable, and for @theacharnian@lemmy.ca I will also add that it is separated into a different file, not hard-coded.
from src/lib/policies/system.ts:
export async function updateSystemFilterPolicy() {
if (SBDisabled && get(userSettings).enableToxicMode) {
SYSTEM_POLICY.set(BLANK_SYSTEM_FILTER_POLICY)
return
}
// auto-update of blocklist continues below
- source
This feature definitely needs greater transparency, and people can debate whether and how offering a “toxicity filter” at all offends them (especially if they are caught up in it), or the way it was implemented, but the reporting on this issue could also definitely stand to be more rooted in facts as opposed to tribal politicking. The fact that the OOP gives extremely strong indications that it was written by an LLM also does not help matters any.
I am not saying that this is a nothing-burger, but I am saying that I see an awful lot of posturing and opinions shared vehemently but exceedingly few people (including those cross-posting widely across the entire Threadiverse) actually opening up and reading the code to look at the facts.
I would think that people would be in a better position to judge, after first-hand reading an actual source of knowledge rather than merely the title and enormously biased summary? (Which has turned out to be incorrect in several places, notably in CRUCIAL details that it has left out - which wouldn’t you know it, sounds just like an LLM to me… - not everyone that I see on the internet is a bot, but then again some text undoubtedly is directly written by a bot.)
Though Lemmy and PieFed’s UIs are both fairly polished.