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 haven't given you a name yet. What should I choose... Puppy, Pogo, Toby, Felix... which do you prefer? Pick a cute one!
I agree with your point of view, but for argument sake, you're actually stating the point for davide in that the higher number of found bugs means more eyeballs on Linux means bugs are found quicker and more often (he said it weirdly or I misunderstood). Simply having little to no QA done on BSD could also explain lower bugs found and reported. I'd be curious whether all these advanced attacks on Intel and AMD processors apply to all kernels or if BSD has been winning those battles. Those are winning arguments for security quality, IMO. Same for those bash and/or openssl vulnerabilities and shit like that that catches the whole world off guard.
But the actual data needed is the number of found bugs vs actual bug but not publicly reported, and that's not possible without an omniscient coding God. Now that would be an interesting movie for geeks.
I was a big BlackBerry fan and they bought QNX for their real time kernel used in mission critical, nuclear type shit. At the time, it was considered small and secure. Then BlackBerry went over the code with all their security and fuzzing tools and found and fixed a shit ton more bugs previously unknown...
I encounter one people using EXWM and trully believe emacs is operation system.
Emacs also have many built-in functionality by default. I dont have idea why people still push another feature on it.
I see my
init.elonly haveand major editing mode like nginx-mode
In any codebase there is this issue, popular ones have more eyeballs, less popular ones have not, therefore the popular ones would have more bugs found which would give the impression of less careful coding practices.
There is also the metric of bugs per 1000 lines of code, as well as other metrics adding to the confusion.
Our problem is that we are overthinking it. Like the doctors which are becoming hypochondriacs. We are aware of the threats and see them everywhere.
It all depends on what you are planning to run and how mission critical it is, how many users it would have and secure accordingly.
CIA/NSA style security, definitely BSD plus tons of ACLs and one-time key generated by a token which could be wiped remotely (or even a disposable token with only 1 usage).
The average joe sharing his copy of a windows game so his buddies could connect and multiplay, just open the port in the router or not even that over IPv6 and done.
Everything else is in between, there is a trade-off between security, availability and ergonomy, you can't have all to the max.
@rcy026 I don’t think it’ll ever be a fair comparison to compare reported bugs on FreeBSD vs Debian, it’s just silly
By that logic, Chrome, Firefox, Edge & Safari must be horrible compared to any 3 starred Github repo of a weird Chromium-fork
It just can’t be fair
BS. That's optional. FreeBSD has nearly all ports also available as packages (ready to install). And btw to build a port is one simple command while building from source is a major pita.
I did plenty testing on both FreeBSD and linux and I don't see that "way better CPU perf on linux".
Obviously @rcy026 was perfectly right.
Re linux vs. FreeBSD, my personal take:
I've learned and come to prefer FreeBSD on servers but linux on desktop/notebook. I perceive FreeBSD as more professional (plus not GPL-tainted which often is a problem for companies) and linux as more "versatile".
One usually doesn't need weird software on servers but I've seen much developer stuff (e.g. in formal verification) that either expressly only supports linux or is easier to install, update, etc. on linux. Similarly linux supports more architectures incl. weird ones and probably also seems to support more graphics stuff (not sure as that's totally not my world or interest but it seems to be like that).
That said those two OSs clearly address quite different people and use cases. Also one must see that linux has way more (albeit also less capable on average) developers and support by large companies. Which btw seems to be a factor most linux fans seem to ignore or not even know about. Clearly, linux is strongly influenced (actually controlled imo) by certain companies, while FreeBSD is not.
TL;DR Both, FreeBSD and linux have their place and their advantages and disadvantages (and I use both of them).
Fully agree.
Influenced somewhat, perhaps, controlled, no way. It is like saying that, because Renault controls a big chunk of the automobile market, then the automobile industry is controlled by Renault. Not even the ICE market, let alone the EV one where everyone has their own idea.
Yes, the general principles are the same, it is a cage of metal (usually), with an engine of some sort, a source of energy and needs some basic environment (roads, but not always) to function with some autonomy. There are some safety rules an regulation, emission standards and the like, but that does not mean any one company can set those standards or enforce their application, every area has different needs and opportunities and the producers are catering to those, the more choice there is, the better. We can have the Fords, the Teslas, the Renaults, Toyotas and many other big producers, but we can also have Neos, Jetcars and Kamaz, all useful in certain conditions since they are still alive.
Nobody controls Linux as nobody controls the Automobile industry. Not even the governments.
I see! It must be like that arithmetical rule they teach at elementary school... if negative times negative gives positive, idiot plus idiot must be genius!
More eyeballs does not always equal lower bug count.
OpenBSD is an excellent example of this. Much smaller dev base. Their culture is hyper-focused on security, they continually audit their code, and they test it on a variety of platforms (I think only NetBSD supports more), and so they rarely have CVEs. Two remote exploits in...something like 20 years. They also have a ton of features (some adopted by Linux much later) to prevent exploits if there is a code problem.
The trade off is a smaller code base and leaving out some features (e.g., Bluetooth). However, you do get a cool song with each release.
It has very cute logo. Way more cutter than Linux
. That blowfish saw things in life that we could only imagine.
I tried FreeBSD 13 while back and it was slower in everything CPU demanding (databases, ZSTD compression, Node.js server) compared to Ubuntu 22.04. Noticeable difference in everything I mentioned, especially ZSTD.
Right now I'm making project where I use AV1 encoding and Clear Linux + SVT-AV1 is leaps ahead of every other Linux distro I've tested. Although I didn't tested FreeBSD I imagine it will be ever worse (Clear Linux is from Intel, SVT-AV1 is from Intel) and AV1 is becoming really huge thing.
Could you please try to actually contribute in any way to the discussion? You have clearly shown already that your statements were completely pulled out of your ass, and these kind of childish outbursts does in no way make you look any better.
You trying to call people idiots after your posts in this thread is just...well, you figure it out. Or don't. I don't even care anymore.
You are correct, but the correlation is not linear.
Once a project reaches enough mass, the amount of eyes should be enough to discover most bugs. There is of course a very big difference between 10 coders going trough the code versus 100 coders going trough it, but the difference is probably not as noticeable with 1 million coders versus 10 million coders.
I would claim that BSD is audited enough to consider it well tested. It's been around even longer than Linux and the code base has always been considered as one of the best around. And don't forget that Apples OS X and major part of Windows uses the same codebase, so it has surely been audited a lot of times.
I agree that in the case of FreeBSD vs Linux it is hard to make a fair comparison since FreeBSD is a complete os and Linux is actually just a kernel. To make it somewhat fair you have to compare FreeBSD vs Ubuntu, Debian, Red Hat or whatever other complete distribution there is, but then you also get all the third party software that ships with that particular distribution.
Im setup and forget guy so for me free/openbsd or slackware. All others are time consuming.
Again, that's not what I see, and I do quite a lot of work with/on DB, compressions, etc.
About the only not insignificant difference I see is that linux seems to offer higher disk performance (due to linux very aggressively caching).
Sounds quite possible to me. While I'm not interested in the "wars" between linux distros, I think it's a credible statement that FreeBSD is weaker in video stuff, in part because afaik quite some new hardware isn't fully supported yet (often exotic hardware support first aims at supporting it at all and only later gets optimized), but again, graphics and video isn't my field of expertise.
Finally, speed/performance isn't everything in real life. In production other factors weigh heavy too and often are even more significant than performance. After all, say adding more cores is easy and cheap, reliability however (usually) is not.
Yep, I also like slackware a lot and sometimes still use it.
Unfortunately that's not how video encoding works.
With video encoding you lose quality if you encode with more cores (Slice-based encoding), because you are limiting how much information can be shared across frames to these slice regions.
More cores = more slices = more useless/redundant information that should be shared across frames instead of being redundant = worse quality at same bitrate.
There is a lot of work put into scene-change based chunking, but it's not an option at all in SVT-AV1 only in some older codecs and it still won't apply to live streaming for example as you cannot look ahead in order to check when there is "scene-change".
If you encode many videos in parallel then ofc you don't care about it, but if you want to achieve best quality real-time on single source (OTT/Live streaming/demanding preset for high quality VoD) then additional performance caused by OS is like encoding on 2026 year CPU
Right now I'm encoding on Ryzen 9 7950X3D + Clear Linux and I can provide sub-1Mbps 1080p livestreams that are really high quality. It's astonishing and I cannot do anything else (other OS, more cores, whatever) because I'll lose this quality edge that I have;)
Nice insight. I am not into video encoding and had no idea. Maybe that is why netflix quality sucks in spite of fast laptop and good network.
You see, I trust that you know what you're talking about - but video encoding is just one niche in the server world.
And btw, if "more cores" isn't helpful, oh well, then more memory or faster memory or faster cores or ... will do the trick.
Anyway this thread is neither about hardware nor about video encoding nor about which linux distro is "the best".
I'll stick to what I said. On servers, except maybe some niches, I'll just continue to not only favour FreeBSD over linux but to in fact not even consider linux. And on my development box and on notebooks I'll continue to use linux.
I summarize it with what I always say: just use whatever gets the job done for you.
There are certain situations and circumstances where there are differences between FreeBSD and Linux but if you are in those kind if situations, chances are you are most likely competent enough to decide on your own what works for you. Very few people push data like Netflix, and people in that kind of position probably knows exactly what they need to look for.
For basic server tasks, go with what you are comfortable with. I've been running BSD for over 30 years and Linux since Slackware was called SLS, and I still don't have a favorite.
There are probably hundred of valid reasons to chose one over the other, and I respect all of them as long as people can argument for them. What I do not tolerate is people just pulling bullshit arguments out of their ass when they clearly have no idea what they are talking about.
If someone were to ask me personally for advice, I would tell them that there are some softwares that do not run on BSD, such as docker just to mention one. And if you really want the latest cutting-edge versions of third party software, Linux is usually faster to implement these. And for workstations and desktops, go with Linux.
But if you just want a well organized, efficient and extremely stable system, use BSD. It might not have all the latest bells and whistles, but everything it does have it runs very very well.
You should always stick to tools that you know the most
If you know how to work with FreeBSD then surely you will manage to do everything.
There is no wrong choice here, I've just pointed out where Linux exceeds in my experience.
With Netflix it's... complicated.
They have their own algo based on psycho-visual data. Basically they gathered bunch of people, showed a lot of scenes and figured out what quality is "good enough" and what determines "good quality" (shadows? lights? banding in sky? skin imperfections?).
https://github.com/Netflix/vmaf
Based on that they are encoding videos to be smallest possible while maintaining "good quality" video. For 90%+ people current quality is on par with traditional media such as BluRay, because they are not pixel-peepers. It's possibly 96%+ at this point, as majority of people consume video only on phones, maybe in home, but maybe on train with shitty network.
So Netflix has three choices:
So this is advanced problem because Netflix is big. Scaling is expensive, shareholders don't like risky ideas. Main selling point is that Netflix works everywhere - TV, 3G network, maybe even your Smart Fridge and its "good enough" quality.
Edit: And now that I'm thinking about it - current strategy gives them option to publish "remasters" in future that will require just a little bit of work (increasing bitrate, maybe pushing it through AI/ML upscaler) and many people will go back to rewatch their favourite video in better quality. Low cost (in future), many customers returning, shareholders happy.
Blu-ray...sheesh, I'm happy with 720p for most things.
And I have 16 bit color depth because xinerama + displaylink cannot handle anything better. Movies kinda look like oil paintings, also at 720p.