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
What do you mean? he said it right there, it's the highest form of honor/award @yoursunny can give anyone.
More consistently reliable, less weird.
Reliable how?
rclone has some quirks, it is not usable for large scale data (lots of random IO). "Endpoint not connected" is a common issue, so you need scripts to check and remount etc. etc.
I only use it as it is definitely the best option for Google Apps/Drive, aside of google driuve filestream which is WIndows/OSX only.
Tricking an OS into thinking a remote store is a local one will include all kinds of quirks when something doesn’t go as planned. Monitoring for various conditions being required long term for sure. I have scripts that attempt to write to remote stores every 5 minutes just to make sure they’re up, because I can’t rely on the OS to report a brief failure in a consistent way.
It’s sort of like building your own imperfect tool for the job, it’s weird because there’s already one made for the role, the reverse proxy.
I'm reading @FAT32 's code and I'm thinking: it's just calling youtube-dl to parse the video info and download the best quality video and audio streams. Then it calls ffmpeg to combine the two streams. Am I missing something ...? On the inspector, I'm only seeing video data being sent over websocket.
(I'm not an expert on this kind of stuff, and I barely know JS)
Thanks, honestly I think this is an interesting technology but I only know very little out of it, let alone understanding how to publish a new object to the network.
Ya that's more or less about it, the thing here is understand / guess how it was encoded / format of the packets at the first place. I also thought about just read from the same WebSocket used in Yoursunny's pushup page and call it a day but that itself have more variables such as adaptive streaming (So it might mix the video in different resolutions)
Truth to be told, I actually have no idea about NDN in the past so well it is also a learning experience for me. There's lots of code in NDNts and sadly I don't see much love / stars in those repos, @yoursunny is working hard to write them.
I first read the code from this 2 repos:
And get to understand that the video is encoded using Shaka Packager as DASH format, but the video can be both vp9 (webm) / h264 (mp4).
Then I take a look at this repo:
To get an understand about what are the packages that we can use, I actually missed out
@ndn/cat
at the beginning that could make the process even simpler.Anyway, the idea of using
youtube-dl
is I am not able to find a good enough node.js library that parse DASH file and download the streams. So I decided to make use of the very powerfulyoutube-dl
that can do the magic for me.youtube-dl
is already sufficient but to merge the audio/video I usedffmpeg
to do the job.SSHFS is really not designed to be high performance. It's a very poor choice for large files or when you need multiple concurrent transfers. It doesn't have any concurrency (meaning only one transfer can be happening at a time), no block access (even if you just want a small part of a file, the entire file needs to be transferred across the network), and no locking. Try NFS instead, or for read-only access you might have better luck with a HTTP proxy (backend server serves the files using HTTP, then some other server proxies them).
Good stuff.
I just found the usage for the library kind of weird, the fact the
face
variable didn't get used right after initialization threw me off. But then I saw the call tofetch
not taking in any parameters which lead me to believe it's using a singleton/global default connection/factory instance that got initialized during theconnectToTestbed
call.Normally, you'd have handles to all the stuff you need to make a call. Just kind of a weird pattern, I haven't read much of the library but I guess you could also manually manage that state if you wanted to.
100% agree, this is what confused me at the beginning also, I was wondering where else can I use the
face
? Why does 1 library affect another? It drives me crazy at firstBut I have to say the TypeScript / JSDoc is pretty well documented.
I see, I mean I haven't really had any issue with SSHFS video serving so far. It currently mounts a local mount to the server and then uses read streams in Node to serve these files. Btw important note, I'm only reading files I won't be writing anything. I did try NFS with read-only but people advised me in LET to not use it remotely because it is not designed to be used publicly, and to only use it within a private/LAN network. I'm not too sure about the HTTP proxy, I ideally want a solution like my existing one (SSHFS Local Mount + Node.js ReadStream Serve from Local Mount).
What? No, SFTP handles partial file operations just fine.
Oh really? Since SSHFS is based on SFTP, I assume even partial file operations are fine 🤔
That's right. If you're accessing a remote server, you can use it over a VPN (eg. WireGuard). https://d.sb/2020/12/nfs-howto. NFS lets you use CacheFS too, to get read caching on your client (if you want to use it).
Oh yeah, it's very possible you're correct, but I'm not sure whether SSHFS fully takes advantage of that. I remember reading on ServerFault that there were some limitations of SSHFS block access (I think it was that SSHFS will always cache the entire file even if you only request a portion of it), but I can't find the post now.
Do note that SSHFS is not actively maintained at the moment:
https://github.com/libfuse/sshfs
SSHFS was a useful for testing the concept. I was suprised by how effective it was in my low volume usage. But yeah, suboptimal long term. After a day of googling and copy pasting, I've updated one of my test PeerTube servers to use rclone as a service to map the drive instead. Seems to add a bunch more options for the off-site storage as well.
Added MergerFS to merge the local recently added folder and remote back catalog folder which should take care of most of the needed caching automatically if the cron prune job works right. Saving money and learning effective tools, what a wonderful site this is
Award delivered.
https://pushups.ndn.today select 2021-06-03 title
Moreover, if the "uncaught promise rejection" refers to a UDP RECVMSG error, it's been fixed in latest nightly.
I never noticed this error because it doesn't occur on Windows.
There's a global
Forwarder
instance in the@ndn/fw
package.Many functions can accept
Forwarder
as an option, with the default being that global instance.I don't know why but it feels so weird to watch someone doing pushups Thanks for the award
By the way I noticed that your push up skill has improved a lot! Shorter interval / resting time and more accurate posture. (I am still very weak on pushups, can only done less than 10)
Thanks for fixing the bug, honestly I wonder why you are so dedicated on working on NDN.
FAT32 is my role model.
Learn from @lentro .
My push-ups are his favorite pastime.
If you are comparing to the 2020-12-24 title, one reason is, that video was the fourth take.
I do not carry a tripod.
To record an outdoor video, I need to place the phone between tree branches or rocks.
Problems occur frequently:
At the fourth take, I've already done 96 push-ups, so that the form is worse.
Indoor videos are easier because I'm familiar with the chairs, and there's no wind to knock off the phone.
Most only need one take.
Several providers have push-up discounts:
You'd start training.
I have a PhD on Named Data Networking.
I skipped the whole thread so someone may have mentioned this:
Encode your url with jwt/paseto. That also means that you have to template your html, so they can't be static. Easy enough to do with php/python/nodejs/etc.
@seriesn 200% recurring discount for each month that we do 1000 push-ups?
In one day yes.
F11.