Privacy is Priceless, but Signal is Expensive
u_tamtam @ u_tamtam @programming.dev Posts 6Comments 454Joined 2 yr. ago
u_tamtam @ u_tamtam @programming.dev
Posts
6
Comments
454
Joined
2 yr. ago
The thing is that I have experience with other complex or high usage PHP applications and I know how to optimize things. What I see in NC is poorly structured code, warnings and erros thrown around left and right.
Well, on my instance, logs are pretty quiet and I am not a PHP developer to form an opinion on the overall architecture. But if you take the time to write down what you feel is wrong with the nextcloud codebase, I'm pretty sure many people (and me first) will read it with interest and perhaps even do something about it (typically the kind of "HN frontpage" content, if it's well written).
The OP also said that ownCloud gave him a much better experience out of the box, and that’s still a “complex” PHP application.
Last I heard of ownCloud, people were saying that it had been rewritten in go or something similar. Funny bit of history, nextcloud forked off of owncloud, got a ton of mindshare in the early days, and quickly became the better/faster of the two (perf was one compelling reason for me to migrate back then). I wouldn't mind NC following suit (in the end, we benefit from this type of competition).
NC webmail is unusable
I don't plan on ever using it, but thanks for the heads-up. That said, if you feel that roundcube performs better, it happens that someone has packaged it for NC, so you should be able to use that instead of the troublemaking client.
Yup, that's a big reason why centralized protocols aren't sustainable. XMPP is 25 years old (which is older than almost anything else on the contemporary internet) and thriving. Unfortunately, judging by the cycle of messengers coming and dying, and people still being eagerly part of that, this isn't something that people value very much.
If you are curious, you should give XMPP a shot, it's equivalent to Signal in terms of encryption, but anyone can host their own. Signal is ideologically opposed to anyone but themselves being in control of your account, and because of that I don't want to trust them.
Yeah sure. I’m not the only complaining as you can read on this post
I'm not saying that you are the only one complaining, but from what I can tell, most people in your situation are deploying their instance from "cookie-cutter" docker images. In practice, it often means that the same machine end-up hosting multiple web servers, database servers, application servers, etc etc. And those servers are developed around heavily-optimized event loops that assume direct access to the full server resources. So if you want predictable and good performance, there's no way around tweaking some knobs and be very mindful of how each and every service is deployed alongside the next one. And of course, you can't trust someone else to know better than yourself what's running in your box (not even the nextcloud developers) and which service should get preference over what under heavy load.
Nextcloud has that against itself that it uses advanced php features and large objects that need to be cached at different layers. That makes it a slightly more complex app than your go-to php CRM, but it's not unheard of either (you'd be at the same spot hosting a large mediawiki or wordpress).
Does that make it garbage? Well, you are entitled to your own opinion, of course.
Also your comment tells me that you’re full of shit, because you’re implying that both a generic Docker setup and mines are all shit. You can’t have it both ways. What are you suggesting? That the NC guys made a bad job out in their Docker images?
Do I deserve the insult? I answered the docker part, though. In general, I'd say that you are better-off not using docker in PROD , unless you have the time and energy to spend rebuilding images to make them fit your pre-existing deployment (what nobody does), and then invest the time in fine-tuning through multiple containers (which very much goes against the "fire and forget" mindset of most docker users).
How many users? How much data?
About a dozen, 2TB, upwards 700k files
Btw do you use the webmail at all or are you about to tell me that these screenshots are hallucinations?
I'm definitely not using the webmail. If you have performance issues, you should rather start with just the "core" (i.e. files) and add on incrementally.
And again, I'm not saying that nextcloud doesn't deserve being optimized, or somehow be made more foolprof. I too went through a phase of "that can't be real, this cannot be that slow" and walked my way out of it.
That might tell us more about how badly your php process manager and/or db connection handler is set up, seriously. I run nextcloud "natively" (no docker, no nonsense) on hardware that was modest 15y ago (Intel Atom/2GB RAM), and it's pretty good.
How isn't that even better, then? Those countries set a bar for themselves, no matter what China decides to do
Does that qualify as a dumb phone though? Symbian could do a lot more, and better, in the area of productivity tools, multitasking, customization and apps management than android/iOS did, and for a very long time. The form factor wasn't putting as much emphasis on the screen real estate but that doesn't make it less smart.
Couldn't you just install nextcloud and none of the apps you don't need? I mean, it's pretty modular..
Yeah, screw them. Let's also not pretend that China is the bar to pass either. Two things can be bad.
Who is "we" exactly? Most developed countries have reduced their emissions so much that they've been producing less CO2 per Capita than China for a very long time (like, a decade for the whole of the EU if I remember correctly).
And I see you coming, no, it's not because China exports lots of stuff for the rest of the world: https://ourworldindata.org/grapher/production-vs-consumption-co2-emissions?country=CHN~OWID_EU27
One can hope! Or maybe that's just the economy slowing down, as anyone walking the streets of Shanghai pre and post COVID could tell.
Yup, and I remember clearly a whole army of plausibly state sponsored shills downplaying/voting the story
It doesn’t really matter if Microsoft/OpenAI are the only ones with the underlying technology as long as the only economically feasible way to deploy the tech at scale is to rely on one of the big 3 cloud providers (Amazon, Google, Microsoft).
Yup, but as the "no moat" link I posted implied, at least for LLMs, it might not be required to spend very much in hardware to be almost as good as ChatGPT, so that's some good news.
Are you saying you’re cool with neofeudalism? Or just agreeing that this is yet another inevitable (albeit lamentable) step towards it?
Oh, crap, no, sorry if I wasn't clear. I believe we are at the crossroads with not much in the middle between our society evolving into extensive interventionism, taxation and wealth redistribution (to support UBI and other schemes for the increasingly large unemployable population) or neufeudalism. I don't want billionaires and warlords to run the place, obviously. And I'm warry about how the redistribution would go with our current politicians and the dominant mindset associating individual merit to wealth and individualistic entrepreneurship.
Sure, now which pre-existing piece of xmpp based software checks all the feature boxes as noted by both Signal adherents and myself regarding Session?
All of those. Essentially you would have to go out of your way looking specifically for incompatible clients.
And "incompatible clients" is simply the natural state of any technology that's been around long-enough. The only way Signal fends itself from this is by mandating its own client and version (and banning anything else, technically or from its ToS) which is terrible for a bunch of reasons (you must agree with Signal's direction and whatever features they might decide to add and remove for your own good, you cannot use Signal on devices/platforms that Signal has no resources/interest to support, etc). If Session is in any way open, and assuming it ever becomes successful, it will face the same challenge (just like Matrix does).
To help you out with the monopolistic/capitalist concern: https://simonwillison.net/2023/May/4/no-moat/
tl;dr: OpenAI's edge with ChatGPT is essentially minor (according to the people from within), and the approach of building ever larger and inflexible models is challenged by (technologically more accessible and available) smaller and more agile models
Imagine a future where most fast food jobs have been replaced by AI-powered kiosks and drive-thrus.
Funny you bring this one up :)
https://marshallbrain.com/manna1
Imagine a future where most customer service jobs have been replaced by AI-powered video chat kiosks. Imagine a future where most artistic commission work is completed by algorithms.
To a large extent, we have been there for a long time:
https://www.youtube.com/watch?v=7Pq-S557XQU
This, and the theory of bullshit jobs:
https://strikemag.org/bullshit-jobs/
were formative reads to me.
The end-game is pretty clear: we have reached the limits to the model on which our current society is built (working jobs to earn money to spend money to live). We now have excess supply of the essential goods to sustain lives and scarcity of jobs at the same time. We will have soon to either accept that working isn't a mean to an end (accept universal basic income and state interventionism), or enter a neofedalism era where resources are so consolidated that the illusion of scarcity can be maintain and justify the current system (which essentially the bullshit-jobs is all about).
It's perhaps the most important societal reform our species will know, and nobody's preparing for it :)
Imagine a future where all the news and advertising you read or watch is generated specifically to appeal to you by algorithms.
This is already the case today:
https://en.wikipedia.org/wiki/Filter_bubble
And this is already weaponized (e.g. TikTok's algorithm trying to steer the youth towards education and science in China and towards … something completely different in the rest of the world).
That’s been your argument this entire time. You kicked around all this time till now saying really weird things like how batteries are inefficient or that green hydrogen is from hydrolysis but then tell me what your point is all along when your point has been wrong from the start.
Let's keep this simple. It all started with your affirmation that energy storage is a solved problem. When I asked how you would go about implementing the solution, you brought-up pumped hydro. And we ended-up with enough data pointing towards this problem being all but solved (cost is one aspect that you are quick to dismiss, but engineering/practicality is a major one).
In all, we agree, we are in the same boat, we want more budget being allocated for the energy transition. But where we diverge I that I don't see how turning a complex problem into a caricature (bordering a conspiracy theory) helps anyone. The physical world we live in doesn't care about opinions, and isn't affected by digital money. You don't have to believe a random stranger on the internet (who happens to work in this field), if this is your crusade, there should be people near you, academics, scientists, engineers, who would be pleased to educate you on the subject. This is pedant, I don't see where's the belligerence.
Not only have I made an argument, I even provided detailed sources supporting it https://lemmy.ml/comment/5942658
What are you talking about? This isn't a response to me nor OP, this isn't even a response you posted in this thread, and your response is no longer even listed on the source article thread. You are just throwing insults faster than you can read, apparently.
Meanwhile, you provided a wikipedia link to autocracy and then said that’s what Chinese system is which is what actual trolling in this thread is.
No, it is not. I am entitled to my opinions, which I substantiated, and you should accept that not everyone agrees with yours. But more importantly, I'm not criticizing you for not agreeing with me (like you do), I'm criticizing that none of your posts contains anything of substance other than insults. You really can't turn that into a problem "with me".
If you cannot tolerate comments about China, from someone who spends a fair amount of their time and life there, I guess you can just block me? Bye.
I really don’t understand the obsession here in comparing energy storage to energy production.
The storage of electricity doesn’t have to meet energy consumption
why, in your opinion, is this more an obsession than "pulling power cables" and "tugging floating wind turbines"? This is very much part of the grid transitioning towards more intermittent (and renewable) energy sources. We can't just keep putting wind and sun without offsetting the intermittence (since we are also removing carbon-heavy sources), which means either adding low CO₂ base-load (nuclear), but we are not going there fast enough, or adding more storage (and neither there do we have a solution).
The first comment I posted shows how if you had 100 the size of the bath county plant you could run the entire US for hours. In just 100 of them. For the cost of the F35 it could be 300 or more but I am accounting for nothing but problems.
It's funny, because my link https://sandia.gov/ess-ssl/gesdb/public/ shows that there are 1693 such projects in the world, with 739 by the USA. China, with a more important landmass and not bothered by F35s (or whatever) doesn't even cross the 100 threshold. So the onus of the proof is on you to demonstrate that we can actually build hundred more pumped storages in the USA for it to make a difference.
From the perspectives of the grid operator, renewables represent risk that destabilizes power delivery. Although weather forecasts are steadily improving and provide more leeway to prepare for sudden changes in the power supplies, the degree to which grid operators can turn on alternative power sources or alert customers to adjust their power demand is limited. In a truly “fossil fuel-free” energy system that relies exclusively on various renewable energy sources, the only viable means of addressing intermittency is to deploy energy storage.
Your source even agrees with me.
This isn't even contentious. What is, is that you believe that we have this silver bullet of pumped hydro to cover our upcoming energy storage needs. And that's not nearly the case.
Once paired and optimized for cost, the model returned 11,769 sites in the contiguous United States, as well as an additional 3,077 sites in Alaska, Hawaii, and Puerto Rico, where closed-loop PSH technology can be best deployed in the future. https://www.energy.gov/eere/water/articles/wpto-studies-find-big-opportunities-expand-pumped-storage-hydropower
Which was my point all along
It is a solved problem. The solution is just extremely difficult and expensive.
I don't want to argue about semantics. If the solution is too costly to be implemented, then it's not a solution. I don't think there's more to be said here.
Don't be too worried about AGI being a thing in the short. And the only thing which I find to suck with respect to consolidation is that contemporary AI requires a lot of hardware thrown at it while cloud services (providing this hardware on demand) are practically the same triopoly. That sucks if you want to be the next AI startup. But academia is mostly unaffected, and far from lagging behind (multiple open source LLMs are compelling alternatives to chatgpt and not benefitting from OpenAI's millions of marketing and hype doesn't make them less valuable)
Edit: Sorry, I responded to the wrong parent.
I don't believe Matrix is better positioned than XMPP to succeed. On a technical aspect, Matrix hasn't managed to stabilize its protocol, and they've been a decade into it. This has resulted in only a single organization being in charge of the protocol, the client and the server implementations. This isn't sound, this isn't sustainable. And now, unsurprisingly, this organization is in a financial crisis, has lost important customers, has no budget secured to maintain its staff in the next years, and recently underwent a major licensing change that we can only interpret as a shift towards an opencore model at the detriment of the regular user.