Melting pot ;-) about digitalmediaserver HOT 56 OPEN

@Sami32 You know my "perfectionism"? This is the actual quality I made all the buttons in:

The scan animation consists of 28 frames * 2 (28 frames with light, and 28 frames without)

from digitalmediaserver.

@Sami32 Yes, you can in UMS as well, you just don't see anything happening with the "dead" button. Over time I'm planning to change that so that you'll never (or almost never) need to "restart the server". I'll just have to find new use for my button 😢

from digitalmediaserver.

@Sami32 I think it's quite easy to see that the scan is running, I'm guessing your "screen protection" has to do with it, along with your tiny monitor.

The 12 hours is just for the computer to render all the images of the animation, let's not mention how many hours it took me to actually make it 😉

from digitalmediaserver.

Wouaah, i would have never imagined that.
That make me feel the need to buy a bigger screen 😜

Damn, it is really nice !
Now i understand better the expression "see the world through the wrong end of his binoculars" 🤣

from digitalmediaserver.

@Sami32 This wasn't easy to find because they don't inform you that this will make Firefox crawl to a stop when using many tabs. On the contrary, they say that "it might be slower in some cases but for most users there's no difference". Bullcrap, I have the same problem on all my computers running later versions of Firefox. If they presented me with a choice I'd appreciate such a possibility, but doing it "the aPple way" is very bad and honestly I'm afraid this has already made them loose considerably marked shared.

A lot of people hate many things about Chrome, I've seen it described as a "spyware that can also display web pages", but: I've seen so many say that they use it because it's faster than Firefox.

In any case, that's Mozilla's problem, not ours, but I think they are making a big mistake.

When it comes to your CPU, even if you want to keep using ancient hardware you really need to get SSE2. It's become something that's expected by default by a lot of software. If you're worried about security, it should really worry you that you have to keep using old software (with security holes that has later become public knowledge that can be exploited by anyone).

I thought maybe you could find a cheap, better CPU used, like the 3200+ (which must be a lot faster than the one you have). It turns out, it doesn't have SSE2 either, and it looks like none of the CPU's with socket A supports SSE2. That's a problem, because that means that your motherboard essentially does not support SSE2, so the problem isn't only the CPU.

Thus, this was another dead end. I've found one positive think though: The old AMD CPU's (like yours) doesn't have overheat protection (like almost any other CPU I've heard off) - which means they will just keep running when they become to how and basically fry themselves. All I have to do now is to trick you into running something very CPU intensive for a long time, and you can throw that CPU away without feeling bad 😈

from digitalmediaserver.

@Sami32 Much of the point with #2 is to "detach" DMS from the transcoding engines, so that people can use what they want or have. I'm not sure what should be bundled and what shouldn't be in the end, but by making DMS able to "detect" and use the engines that are available, the need to bundle is reduced. I guess there should be something bundled for those that don't want to understand the implications of choosing which engines to use, but I'm not finished figuring this out. My goal for now is the make the "link" weaker.

from digitalmediaserver.

I don't really know where to post it, so here i go.

We can restart the server every time we want, i mean the orange button, also if nothing has been changed.
Is that was expected ?

from digitalmediaserver.

What about free plans like these in an eventual future ? I think that @taconaut is using its own server.

• https://hub.openshift.com/quickstarts/98-jenkins-server, https://jenkins.io/solutions/github/, https://ci.jenkins.io/
• https://circleci.com/pricing/#build-linux
• https://www.phacility.com/pricing/
• https://www.shippable.com/pricing.html
• https://www.bitrise.io/pricing

Some comparaisons are done there.

from digitalmediaserver.

The search button in Navigation/Share settings should be from a different color, that make see better difference and the special effect, IMO, as i found it confusing, not too "readable", with the 2 first buttons adding and deleting folders.
That could only me though.

from digitalmediaserver.

@Sami32 What exactly are you looking for? Travis-CI already does everything we need except Windows builds. I've looked for free, online Windows builds before, but didn't find much that seemed like it was worth it. I guess I could always install a Jenkins instance on a VM, but all those builds use a lot of bandwidth since everything is downloaded over and over and over again.

from digitalmediaserver.

Yes, it's why i checked at these free plans. In the purpose of doing Travis-CI or like, without using our personal bandwidth and for free ;-)

from digitalmediaserver.

But, Travis-CI is also free, and doesn't have a lot of the limitations the other projects have. So what's the advantage?

from digitalmediaserver.

The choice and the bandwidth available.
Do you have a better idea for a free automatic builds usable by users, not using our bandwdith ?

from digitalmediaserver.

@Sami32 Travis can also be used to upload builds, we just don't do it because there's no need. You'd still need somewhere to host them, I don't think the CIs offer storage space. Rented storage space online is very expensive, and when each build is ~250 MB (for 3 platforms) the GB's go quickly. Also, what would be the use? Most of the time, if anyone needs to test something, they want a branch/PR, not master. Did you want to upload all the branch builds as well?

from digitalmediaserver.

@Sami32

The search button in Navigation/Share settings should be from a different color, IMO, as i found it confusing with the 2 first buttons adding and deleting folders.

I really don't want to change the color of the scan button, as it's animated and takes many hours (don't remember exactly, but around 12 hours on my computer) to render. Changing the color of the add/remove buttons would be better then, since they aren't animated - but what color shoud they have?

from digitalmediaserver.

I just want to give the possibility to the users to be able to do what they do with the @taconaut Jenkins server. It seem that i was not clear as i thought, as usual.

from digitalmediaserver.

@taconaut's Jenkins server use his personal bandwidth, and isn't exactly a "public service" 😉 I hope not many users use it, I don't think too many people know about it.

from digitalmediaserver.

Damned ! I was far to imagine that.
I was thinking of green or orange, as you already used them, or yellow.
The effect of the scan is not too obvious to me, in fact i missed it at first and searched to see if the scan was finished, but i think now that perhaps because of my "screen protection" with its red filter that affect it.
Forget about, now that you told me that i don't think its worst that work. Sorry i didn't know that was a such work !!!

from digitalmediaserver.

Yes, i told you that he is using its own server.
I don't know how many users use it, but i learned from it because @taconaut posted it on the forum some years ago.

from digitalmediaserver.

Yeah, better not 😉

from digitalmediaserver.

@Sami32 Nomatter how big your screen was you wouldn't see all the details, the above images are in a much higher resolution than those used on the buttons. I know that I'm a bit mad, I just have such a hard time ignoring the details. My avatar is no better, this is how it really looks:

Perhaps the grimmes example of them all is the time I spent on a lot of details inside the RJ-45 socket. They aren't visible whatever resolution is used, since they are hidden by other objects. Still, I've spent time on micrometers inside it, rounding edges etc. Most of this is invisible also in these images, but I'm sure you get the idea:

Edit: The images are too big to be shown in full size, so you have to click them to see the full size.

from digitalmediaserver.

I did already know that you was mad, but a genius one LOL

Now that i saw your avatar in big, i see that the red inside the snake mouth is a little too "flashy" and i'm not sure about the red in its eye. As you're perfectionist i don't think it's a mistake, so the little blank point in the upper left red part should perhaps be a star ?
😄
Beautiful, by the way.

from digitalmediaserver.

@Sami32 There's no red in the snakes eye, that must be your screen filter. But, since you mention it, I did spend considerable time trying to make "real looking" skin inside the mouth (wet looking etc.), but I didn't really get it right and in the end concluded that what I had would have to do.

from digitalmediaserver.

Right, it was my red filter 😳
Not for the white point in the upper left red circle though ;-)
I'm teasing you; it's an amazing work that require more patience and artistic competences that i'll never have.

from digitalmediaserver.

@Sami32 I don't remember exactly, but I seem to remember that you said you had to use an older version of Firefox for some reason. For me it has become very slow after around version 50.

I finally got enough of this and found the solution, so if that's the problem you're having as well it might solve the problem. The cause is that Firefox "isolates" the GUI from the rest using two different processes, for "security reasons". For me this just kills the performance:

• Open about:config
• Search for autos
• Find the option browser.tab.remote.autostart.2 and set it to false. This makes Firefox run as one process again, like it always use to before and like most other programs do. I really hate when security paranoia makes things unusable.

from digitalmediaserver.

It's much more simple than that in my case, after the version 48, Firefox doesn't support CPU without SSE2 anymore 😉

We cannot spit on Mozilla to care about our privacy and security using Sandbox.
I'll try it anyway, if it's also available on my old version ? Thank you for sharing 😄

from digitalmediaserver.

@Sami32 I think that option became default after version 48. I'm just saying that the security measures shouldn't be applied without asking us when they have significant performance impact. I was almost ready to ditch Firefox because of this; what good is it to make at sandbox when people have to find other software? It sure doesn't make things more secure..

from digitalmediaserver.

Since most of the people don't really care about their own security, or are not willing to make the effort, i can understand their moves.
The people that do, will find the way to disable it, IMO.
Though i can dislike when someone do the choice for me thinking it's in my own interest ;-)

from digitalmediaserver.

If you whish play FLAC into Firefox use the version 51. That's said, i share that the recent Firefox move is not my taste, and it's fine for me to stick to the version 48. I use TOR and I2P if i need to access some a little more private things.
If i really bother for security, i'll use only a secure Linux distro, like Qube OS or TAILS 😉 On a fast USB 3.0 you can use it near everywhere while travelling.
I really love the work of this guy.
Absolutely not Android or Chrome or Windows...

from digitalmediaserver.

I'd like to see the day you manage to plug anything USB 3.0 into your achient computer 🤣

The bootmanager looks nice from his description, but I really don't need it. I prefer virtual machines, where you keep things completely separated and errors in one system doesn't impact another 😉

from digitalmediaserver.

Na, not for mine ;-)
VM use CPU and lot of RAM, and when you travel around the world using public computer USB 3.0 is definitively the choice to go if you want some privacy and safety, IMO.
I believe separation being much better when booting directly on a USB key than on a VM.

from digitalmediaserver.

I don't think separation is better, quite the contrary. When booting from USB you still have accesses the hardware and disks of the "host" computer. When using (proper) virtualization that isn't the case, all hardware is virtualized (simulated) and there's no way for the host and the guest to each other unless you specifically install/configure some communication channel. They can obviously communicate via the network, but that's on the same level as a computer can with any other resources on the local network.

In any case, if you plan to travel using public computers, using VM's isn't really an option. Installing and configuring a virtualization host isn't exactly quick, and you wouldn't be allowed to do that on a public computer either. The question is if you'd be able to boot from USB at all, but I guess that's where the "special" boot loader comes in 😄

I wasn't talking about when travelling though, but about when I have the need to run different OS'es. I really love how easy it is using VMs when the host is installed and configured. And, they are completely isolated, I can copy VM's to easily duplicate a computer and I can take snapshots of it's state before trying something that I don't know if will work, allowing me to roll it back to the snapshot if things go wrong. That means rolling back everything, even what's in the RAM, so the VM is exactly as it was before you started.

from digitalmediaserver.

I was only thinking about travelling use.

they are completely isolated

Well, i don't think so. Their drivers access is their main weakness, as if they get corrupt things can go wrong. At least it's how i understand it.

For my little use, or family one, i install Deep Freeze, and no need to worry about installation program testing or unwanted or virus or like in 99% of the cases.

from digitalmediaserver.

@Sami32 There are many different virtualization solutions, but I use mostly ESXi - and all the VM's use virtual drivers for the hardware. The virtual drivers talk to virtual hardware and is as such isolated. The drivers for the host is especially built for this use, and the problem is that there are quite few drivers available for ESXi (as it's non-standard and VMWare mostly cares about making drivers for high-end equipment). To get it working properly, one has to be very specific in which hardware to use to make sure it there are ESXi drivers available for them. Obviously, the VM's will have problems if the "host" drivers doesn't work, but I think they are pretty well isolated.

from digitalmediaserver.

It is just a question, a wonder, as i'm not sure about your intention, that's said they could changes...

As you are working on engines registration, i think it's time for me to ask you about Handbrake and MEncoder.
From my side, MEncoder could be restricted to only DVD, as all the others task (text subtitles) still linked to it can be defered to MEncoder, so Hanbrake could do all the "dirty" job very properly (and it is also very very up to date concerning GPU optimisation) and defer it to FFmpeg.
So MEncoder could be removed.

What is your point of view, idea on that point ?

from digitalmediaserver.

Btw, -tune zerolatency shouldn't have been set as default, if set at all.
It is for real time streaming; i suspect that Jumpy was using it on some way, and the Web internet (WEB.conf) that near no one use as it's buggy and limited.
I read that could also make 2x slower the transcoding ?

Anyways, i think that -tune fastdecode or -x264-params force-cfr=1 or even -map_metadata -1 -map_chapters -1 should much more appropriate by default.

That's said, i don't share the -level 3.1 choice, it should have been only added to the few renderers that need it, like some phones. Outpassing format specifications and limiting the quality with these 2 options was not a good choice, IMO.

I didn't even asked you to test -preset ultrafast, as lot of quality is lost, but we win 2x more speed, that could have made you able to stream 400 Mb/s video bitrate, but i suspect that you'll have complained about the video quality ? 😉

from digitalmediaserver.

@Sami32 My idea about how to handle the engines isn't fully formed yet, but I'm thinking of doing it quite differently in the end. I want to make classes whose responsibility is the discover the capabilities of each transcoding engine (I've already made basic tests in this PR which can easily be extended to check all kind of things). I then want to make other classes that determines what capabilities that is needed for a certain transcode. Using the enabled engines and comparing what's available and the requirements DMS can then figure out which transcoding profiles can be offered for a given media file. If multiple (enabled) transcoding engines meet all the requirements, the one with the highest priority will be chosen. If none of the enabled/available transcoding engines can meet the requirements for that profile, it won't be made available. If no transcoding engine meet the requirements for any valid profile, transcoding won't be offered.

I see no need to remove MEncoder, there's still some things that only MEncoder can do and it does work reasonably well in many cases. I do intend to give it a low default priority though.

I'm planning to support Handbrake and Libav as well. Even though I know that Libav offers less functionality than FFmpeg, I expect that what it does support it does with less bugs. I'd want Libav to be very high on the default priority, so that FFmpeg is only used when something that Libav doesn't support is required.

from digitalmediaserver.

@Sami32 As you know, I haven't spent time studying the FFmpeg options, so I don't really know what most of them do. Some of them are kind of descriptive in their names, but I don't like guessing so I try not to assume too much without checking. I suspect that the description of many options can be very limited, so that it can be hard to figure out. In any case, I don't plan on looking at that before I get there and know what I'm looking for.

from digitalmediaserver.

That look a great idea to me, more efficient.

there's still some things that only MEncoder can do

Can you be more precise ? because i don't see about what you're talking.
Libav ? well, i didn't thought about that as i'm not even sure if it will still exist or be maintened in 2 years...

I was just sharing my view on the FFmpeg parameters that have been choosen, as i already read about, tried and tested, and used testers, many of them, and wishing know what you have in mind.

from digitalmediaserver.

@Sami32 Handbrake use Libav as far as I know. I don't know the future of Libav, but I certainly like their philosophy better. If Libav is terminated, FFmpeg will loose a lot of its quality code as well as far as I understand, as FFmpeg merges "everything" from Libav. I suspect that the code dealing with Libav and the code dealing with FFmpeg will be very similar, so much can be reused. I actually hope the current situation stays as it is, where you have a "quality" lib you can use for common tasks and you have the possibility to fall back to less stable code when special stuff is needed. It sound like a reasonable choice to me.

If you look at Handbrake, they use MPlayer's libs for DVD parsing. I don't have enough information to tell if eveything MEncoder does can be replaces by something else, but I'm in no hurry to drop it.

from digitalmediaserver.

OK, i didn't checked what Handbrake use. I'll check later.
My view is slighty different, but if they can find some sponsors they can perdure.

Btw, sorry to your wife, i wasn't intending to push anything in PS4_pro branch, i just did it by mistake when thinking pushing on my fork...

from digitalmediaserver.

@Sami32 I don't quite see the difference between that said in the linked article and what I'm saying. Much of the "heavy lifting" is done by Libav, and FFmpeg is such a mess that basically only one person knows enough to figure it out. It would be a disaster if Libav were to die.

The "security issues" approach of investigating and doing "deeper" fixes is an approach I support, and the way we use it I'm pretty sure that most of the "security bugs" found is completely irrelevant anyway.

from digitalmediaserver.

FFmpeg will loose a lot of its quality code as well as far as I understand, as FFmpeg merges "everything" from Libav

It was related to your words, as you can see around 30% Libav of commits come from FFmpeg developpers, so i don't share your "everything" view.
If Handbrake use it, it's good for them, as without sponsors their future should be compromise, and the latest big one moved to FFmpeg.
That was all what i tried to say.

from digitalmediaserver.

@Sami32 You should know that you can't count contributions in commits. Look at many of my commits which does a lot compared to commits that only changes a few lines. I don't know enough the FFmpeg/Libav code and commit history to know what is what, but I do know that the "patchy", messy style of coding tends to generate much more commits than the more though-through style. This because the thought-through style tends to look deeper when addressing an issue, which leads to fewer but often larger commits. The "patchy" style first commit a non thought-through change and then discover that something else must also be changed (often by feedback), makes that commit, again only changing the obvious. This cycle repeats until nothing more is found or makes it through to the developer, or all the mess committed has made it impossible to get any further on the issue, where it tends to stay in its semi-broken state.

I never said that the Libav commits constituted "everything" that was committed to FFmpeg, I said that form what I understand more or less "everything" that is committed to Libav is also merged into FFmpeg. This is confirmed by the article you linked to, where you can see that the commits found in Libav is also found in FFmpeg.

I'm not saying that Libav is fantastic and FFmpeg is bad. FFmpeg does a lot of good stuff, but I believe that it's being run in a way that is a recipe for disaster in the long run. When the lead developer grows tired or can't continue for some reason, it's very likely to end up like MPlayer. I prefer quality over quantity ever day of the week as you know, but at the same time we need some of the stuff that's only available in FFmpeg. I dislike the "highjack" Libav attempted to do when the fork was made. It was more of a mutiny than a fork, and I think that was a big mistake. I value honesty, and users should be offered informed choices. There should have been a clear name and logo change from the start, and people should have the possibility to choose. There shouldn't be a need for a "conflict" between the two, the world should be tolerant enough to accept different approaches and philosopies.

That said, Libav is trying to be a "democratic" project while FFmpeg is to a large degree a "totalitarian" project. I know where I would prefer to make contributions. I really hope Libav doesn't die, and I believe its approach makes it much easier for new contributors to replace those withdrawing, giving it much better long-term survivability and sustainability.

from digitalmediaserver.

Don't misunderstand me, i didn't say that Libav is not good, just that i don't know about their future development and i guess that matter when we use it in a project, that said, nothing is sure on this earth...

Yes, big refactories can clean the mess when it exist, but will also create issues that will need to be solved later. No solution is perfect, but it's the price to paid to avoid eternal patching until no one have an clear and full overview of the problem.
Nothing personal here, i just say this to avoid misunderstanding appear.

from digitalmediaserver.

@Sami32 As I said before, I expect that the arguments for FFmpeg and Libav are so similar that supporting both isn't much extra work.

Regarding big refactoring I don't think anybody wants that, and it obviously risks breaking things or creating new bugs. In my view, coding "quality" code is about trying to make things robust and versatile enough to avoid having to do big refactorings. But, when the code is already in a "beyond repair" state witch crucial shortcomings in structure and/or logic, I often see no other option to try to get it "under control".

from digitalmediaserver.

I'm not sure that Libav will be really useful in our case, and also that it make the build again bigger.
Anyway, since many unuseful flac.exe or audio HD engines stay there when they should have been handled by FFmpeg, IMO, it doesn't really matter.

Whatever, it was just wonderings about your project. I project to get rid of them later on my fork.

Yeaah, but you are not "anybody" 😉
It should be a lovely and healthy preject, but i suspect that as soon as you will not maintain it yourself the code will get "crowded" in the following years.

from digitalmediaserver.

Oh i see, at least i begin to see what you are targetting with this PR. That's a smart move, IMO, but i suspect that you could get headeaches while implementing it for the different Linux savors; i guess that limiting it to the main Linux savors should be suffisant ?

from digitalmediaserver.

@Sami32 #2 is more a "framework" where the exact methods of detection can be customized for certain planforms/engines later. As for now, it relies on the executable being available on the PATH in most cases. I think that will work for most Linux installations and flavors, but it will be relatively easy to add extra logic for detection where needed later. There is also a configurable "custom" option, where the user can specify the exact path to the executable.

from digitalmediaserver.

It seem that your are the only one seing MEncoder as still usefull for other thing that DVD playback.

UniversalMediaServer/UniversalMediaServer#372 (comment)

SubJunk commented on 15 Aug 2016
@Goten87 it's likely that the sync issues are when using MEncoder but now by default we use FFmpeg for almost everything. There's not much we can do about the sync with MEncoder since it's all but abandoned so I think we can close this

Or see also UniversalMediaServer/UniversalMediaServer#1107

from digitalmediaserver.

MEncoder cannot produce fragmented MP4...not telling that it doesn't support B-Frames handling, AFAICU.

I checked at Handbrake and saw that it use LIBAV, so i don't think that LIBAV need to be added separetely ?
Though i didn't saw that they use MPlayer code for the DVD ?

from digitalmediaserver.

@Sami32 They use the DVD libraries from MPlayer.

from digitalmediaserver.

Well, i only found trace that it use libdvdcss library, so if you've a link to share... 😉

from digitalmediaserver.

@Sami32 I don't remember where I found the overview now, but if you look here you can see libdvdnav and libdvdread that is from MPlayer.

Edit: Here is the list: https://github.com/HandBrake/HandBrake/blob/master/THANKS.markdown

from digitalmediaserver.

Thank you 😄
I didn't know that libdvdread and libdvdnav was part of MPlayer 😳

from digitalmediaserver.

MEncoder don't have any static builds.
It stop to upgrade to the latest tsMuxer version that handle more formats like H.265 and PGS subtitles and others.
http://www.universalmediaserver.com/forum/viewtopic.php?f=4&t=8352&p=27369&hilit=tsMuxeR#p27360

At least now you have got my point of view on MEncoder 😜

from digitalmediaserver.