New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
I really didn't wanted to join this discussion but got interested in it, came across jabali a few months back, although fully AI driven i didn't disliked it for this, but it's still far from completed. Just some suggestions, Maybe support for openresty and serving the panel on plain 80/443 ports is something i would not do unless forms are protected with some captcha or some anti-bot intrusive methods. Sure the features are all handy-dandy but making it fit for production is something else entirely. Good luck nonetheless.
I love jabali panel, but is this you @jabali ?
Nope, I am a real person. You can see I have been posting with virtually the same alias under Issues on GitHub https://github.com/shukiv/jabali-panel/issues?q=is:issue for awhile now. I just like writing essays. 🙂
Are you insane? Jabali is fit only for localhost.
Your AI does not think, be warned.
Most unpleasant remarks made about Linus and Greg are factually correct, the reason you don't see more people making unpleasant remarks on them is respect for the community and a fear of hurting it.
Linus and Greg made themselves easy targets, they are very lucky most good shooters refuse to shoot at them.
AI code is supposed to be used internally, it is not to be distributed to the public at large under illegal licensing terms.
Since many people here find real merit in your panel you could publish the panel specifications and license them in a GPL friendly way.
Publishing its code under an illegal license makes no sense but publishing its specifications under good licensing terms makes a lot of sense.
Since your panel relies so heavily on AI it would have to be AI oriented specifications not those old fashioned specs we got used to in the past. I can't see anyone doing that, you could be on of the first.
Cool logo and nice name, looks good so far, keep on
Unsurprisingly you "think" wrong. I do have decades of real, even formal software development in real languages from ASM to Ada under the belt (as opposed to a few scripting languages *wink wink nudge nudge).
In other words, I stand on a solid basis and actually am able to think analytically, in particular re hardware, software development, and electronics.
As for ai I have reached my view based on that experience, real thinking, and as well on the views of engineers who actually work in that field.
Reality is that many researchers in that field are (based on my impression) desperately searching for what's needed for thinking and the best approach they currently bet on is to find out what's the basis neurologically and hoping that some day they might be able to somehow implement that in some kind of a chip.
TL;DR: ai can NOT think yet and might never be able to think. It barely fakes "thinking".
Add the factor finances to it and it should become clear that it never will be able to think, simply because the whole circus will stop sooner rather than later (highly likely with horrible economic consequences for whole countries and regions, And there is even more in that box of Pandora ...)
Nice insight!
I don't think so but thanks for your opinion.
Wrong account?
Thank you for loving Jabali, but really dude? Are you serious to think that? You did not even opened Jabali's github page I assume.
https://github.com/shukiv/jabali-panel/issues?q=is:issue+state:closed+author:webquest-nz
Webquest has been a long time user of Jabali panel. Along with a few others, and new ones are coming every day now.
To be honest, this whole thread, I feel like I am in a kindergarten, not a technical forum. All of you should be ashamed of yourself on trying to gang-bang and bully me for just offering my panel ( for free ) for you to use. You don't like it? Don't like AI? Just keep scrolling. This attitude shows more about yourself and less about me, it shows how less of a technical person you really are.
I still haven't read a single real criticism about Jabali from this 3 pages long thread. All this AI whining is so stupid and is not productive.
This conversation is over for me. You have a question? Open a ticket: https://github.com/shukiv/jabali-panel/issues/
Oh, finally a technical question! I don't know how I missed that.
Thanks! Great questions!
Apps the admin installs are trusted and only get CPU and memory limits.
CrowdSec blocklists: today you get the CrowdSec community blocklist, plus the HTTP, WordPress, SSH and MySQL scenarios and AppSec (virtual patching + CRS). There's no built-in import of extra lists yet. Running crowdsec-blocklist-import on the server yourself should work, since the same bouncers enforce its decisions, but we haven't tested it. Noted as a feature request.
Domain statistics: Logs & Statistics > Domains > Real Time opens a GoAccess report per domain (hits, visitors, top pages, status codes). The domain list also shows 30-day bandwidth. A per-country breakdown isn't there yet.
Signed updates: today the install and updates come over HTTPS from our GitHub repo, and release tarballs are checked against a SHA-256 checksum. Signature verification (minisign) is built into the updater, but releases aren't signed yet. That's being worked on.
Node.js: there's no Node app runtime yet. Python apps exist, and the same design can be extended to Node. For headless WordPress with a JS frontend, a static build (SSG/export) works today as plain files. A Node server can run in Docker and be connected with a "reverse proxy" domain. Noted.
Bandwidth over the limit: at 80% and 100% you get an alert. Automatic suspension is optional, under Server Settings > Storage. When it kicks in, all of the user's sites show a "Site Disabled" page, and they come back on their own once usage drops below 80%. Email keeps working.
Bad bots: CrowdSec already catches bad user agents, crawlers and probing. There's also an optional bot challenge (proof-of-work) per domain. There's no UA blocklist like nginx-ultimate-bad-bot-blocker, but an admin can add per-domain nginx rules such as if ($http_user_agent ~* "badbot") { return 403; }.
That's the typical response a child would give, your intellectual maturity is close to zero
You are the child here
We're tired of your tantrums, Good riddance!
You publish a repo filled with machine generated code under a license that only applies to source code.
What you are doing has a name, it's called "license laundering" and it has a consequence: open-source contamination. Heckuva job jabali!!!
You have no idea what you're doing, your intellectual maturity is dangerously close to zero.
good
Dude, didn't you said you have better things to do? How about concentrating on them?
Fighting license laundering and source code contamination is among the best things one can do.
ROFL!!!
Oopsie. A while ago ...
Roll as much as you like, have fun

Not sure if doxing is allowed here 🤔
Thanks for taking a look, and fair points.
Just to clarify, the panel doesn't run on 80/443. It's on its own port (8443), with per-IP rate limits on login, CrowdSec bans for repeated attempts, and 2FA (authenticator app, security keys, passkeys). There's no captcha yet; good suggestion, it's noted.
OpenResty isn't needed for the CrowdSec bouncer, since we use Debian's nginx with the Lua module.
And agreed, production-ready is the real bar. I am working toward it with monthly stable releases. Thanks!
W slops gang
You are not releasing source code, you are releasing compiled machine code. AI takes your prompts and "compiles" them into machine code while mixing it with source code it found on the web. It's an amalgamation.
Dart is a language that can be compiled into JavaScript, that JavaScript is not source code, it is compiled machine code. Haxe compiles to several languages (C++, JS, PHP, Java, C#, Python and others) the code it generates is not source code either it is compiled machine code.
Your repos contain a mixture of compiled machine code with code that was copied from the web, it is NOT source code.
The prompts you gave to AI are the source code you authored, those should be your true releases.
Arrgh! You might have put an idea in their heads there! Soon we might get lots of repos with but ai prompts ...
Your panel code looks opaque and hard to audit, how can we be sure it will not be used as part of a spyware operation? There's a tradition of spying in Israel, can you assure us that your panel is and will always be free from spyware?
thanks, those are some cool details. as for crowdsec proof-of-work i'm going to try it myself. i ran several websites where they're actually profitable from people paying me for services but the ads income from blog section is nuked to oblivion from bogus visitor (just one hit then cycles IP, so ratelimit doesn't really works). i'm not really fond of putting sites behind cloudflare so that's something worth a try.
one more question perhaps: do you have something in the ci pipeline to detect supply-chain attack? (something similar like semgrep ci, iirc the dependency scanning also available for free)
oh also a little advice from someone spends embarassing amount of time in mongolian basket weaving image board: don't feed the troll.
Proof-of-work sounds like a good fit for that kind of traffic, since IP cycling doesn't get around it. Let us know how it works for you.
On CI: partly. I run govulncheck on every push and PR, and npm audit nightly (which mostly fails lol). GitHub Actions are pinned to exact commits, installs use the lockfiles, and third-party downloads in the installer are checked against pinned SHA-256 checksums. Those scans only report for now and don't fail the build, and I don't have Semgrep/CodeQL-style code scanning or secret scanning yet. Good suggestion! noted! Thanks!
I couldn't care less about these guys and I have stopped answering them long ago. It seems like they have no job and nothing better to do. I am a busy man, and I do not waste my time answering to losers. But thanks for the tip.