(I usually don't post about mkvtoolnix here on Hydrogenaudio, but with CD-in-mka becoming more and more popular I guess this thread might be interesting to you guys. So here we go.)

no, 1.0 has not been released yet

Some of you have probably wondered what will happen with mkvtoolnix after 0.9.9. Will there be 0.9.9.1, or 0.10.1 or what? No. After 0.9.9 I'll release 1.0. Maybe even before that.

So what are my plans for 1.0? Is it 'just another release'? Certainly not. Back when I started mkvtoolnix in January 2003 I had some goals for a 1.0 release. Those goals were few, but some of them were (too?) ambitious. Amongst them were support for the 'usual' container formats like reading AVI and Ogg/OGM, for some raw audio formats like MP3, AC3 and AAC and at least some text subtitle formats like SRT. Also I wanted to support splitting and concatenation.

Some of those goals I've met very quickly (AVI, MP3 and AC3 support - I've basically copied them from my old ogmtools), and some I haven't implemented by now (concatenation). Some are very mature (e.g. OGM / MP3 handling), and some are rudimentary (splitting). But in many, many ways I've surpassed my expections by far. Some examples for features that weren't even on my _radar_ when I started mkvtoolnix were VobSubs, RealMedia reading/writing, support for reading chapters/tags in XML format, having a GUI for my tools and especially _running mkvtoolnix on Windows_. As a matter of fact I'm pretty proud of what I've achieved, and the amount of feedback I get tells me that I've done more or less OK


Earlier I've said that concatenation was part of my 'things to do before I release 1.0'. What about it now? Well, there's AVIMux GUI which can concatenate files, and at the moment I simply lack the time to implement it properly. It will definitely come. That's a promise. But not before the 1.0 release.

For 1.0 I have another feature that I've been working on together with help from Goldenear and Toff: sample precision for timestamps and durations. This is especially useful for audio files. Goldenear is a big fan of making a CD rip and storing everything from the CUE sheet information and the actual contents (compressed losslessly, of course) to CD covers in one Matroska audio file. And everything is supposed to be reversible! So we needed sample precision for timestamps.

I've implemented that. And this is my 'final' feature enhancement before the 1.0 release. The next release will be 0.9.5. It will feature the final implementation of the automatic 'CUE sheet to chapters/tags and back' conversions and it will feature sample precise timestamps and durations.

After 0.9.5 there will be a feature freeze, and I'll use the releases 0.9.6-0.9.9 for bug fixes only. The goal is to have a mkvtoolnix 1.0 package that is as bug free as I can make it. After the 1.0 release the 1.0 line will stay a 'stable' release if you want to call it that - only bug fixes will be included in 1.0.x and 1.1.x. The next release with new features will be 1.2. It might brake more often Well, by now you should know how good or bad I code .

Before I release 0.9.5 we have to make sure that some other apps are fixed because mkvtoolnix now uses 64bit floats in Matroska files, and at least two authors have only implemented 32bit floats in their applications. mkvtoolnix does not violate the Matroska specs in any way. 64bit floats is just a feature those two persons haven't implemented yet - probably because until now no application was using
those.

Anyway. My plan is to release 0.9.5 sometime next week. After that I'll see how the bug reports roll in about the sample precise timestamps and the other stuff. There will probably be a 0.9.6 release, maybe even 0.9.7, but I guess that 0.9.8 and 0.9.9 won't happen - and 1.0 will be released in about a month.

Now you know. 1.0 will be something special At least for me it is. During the last 1 1/2 years I've spent tons of time on this project and everything around it. Obviously it's very important to me. And I wouldn't have come this far if there hadn't been you guys & gals. Each feature request, each bug report, each 'thank you' shows me that my work is not in vain. It keeps me working to improve things. If it were an application just for me I would have stopped long ago.

I hope you agree with my plans for 1.0. If you have some valid objections ('you have to fix this bug I haven't reported' or 'this feature is an absolute MUST HAVE') don't hesitate to speak up

Thanks.

Mosu

... i bet there is a not-so-small number of people here who will test this new feature extensively ... at least this is what i hope ....

Great work Mosu, where would the project be without you and mkvtoolnix ?

I totally agree with your plan for the mkvtoolnix v 1.0...


Yes...you and Matroska Team have made a lot of stuff since last year. So, thank you very much for that.

Well...I've a request concerning Audio formats because we are in Audio Forum here : Is there a plan for the future of mka container ?

Another lossless formats as Wavpack and Monkey's Audio will be great I think. Most people use these formats to archiving losslessly nowadays. The sources are available now and there are portable for multiplateforms...So what do you think about that ??


Anyway, I've another question about Ralf (Real Audio Lossless) but I'll post in the Doom9 forum.

WavPack is on my TODO list. As a matter of fact I've moved all my TODO items from my TODO list to my bug tracking system. You'll find both known issues and the features I plan to implement after 1.0 here: https://www.bunkus.org/anthill/buglist.php?type=unres

'enhancement' means it's a feature request/wish

About Monkey's Audio: It has been requested often enough, but last time I checked its license was totally incompatible with the GPL that mkvtoolnix is released under. So unless somebody writes a (L)GPLed bitstream parser for Monkey's Audio I won't implement it. Sorry.

ok...many thanks for clarifications


Another request (maybe OT sorry...I can't access in Doom9 forum now to verify) : demux vobsub files from mkv via mkvextract. Is there possible ?

Thanks.

It's not implemented, but I'll add it to the feature requests (it has been requested before).

hi mosu, is it possible that instead of skipping bad audio data and losing sync between audio and video when muxing corrupted audio streams in matroska could mkvtoolnix just replace this data with silence?

a big thx for all the hard work you've done in mkvtoolnix

Thanks for you fantastic tool Mosu, mkvtoolnix is really what made Matroska a reality for me.

( Concatenation of files is big in my wish list :) )

Thanks for all your work with mkvtoolnix. I especially appreciate your work on vobsub, flac, and precise timestamps.

This totally depends on the audio codec. I could implement that for AC3 because AC3 is CBR. But for MP3 this is practically impossible if the stream is VBR. How should mkvmerge know how many samples of silence it should replace 300 bytes of garbage with?

Well, broken AC3 happens pretty often. I don't own many DVDs, but on three of them the AC3 tracks for each language (!) are broken in different places, so this might in fact come in handy. But still... Replacing 300 bytes of garbage with silence does not mean that A/V sync will be kept (meaning there's no guarantee that it'll work).

That's one valid point in favour of containers with timecodes

Keep the impressive job Mosu !

I'm looking forward to 1.0. Thanks for the hard work so far and I must say I think your coding abilities are at least _impressive_.

Cheers,

tiki4

P.S. I tested some of the latest builds and I think something has to be updated in foo_matroska as well (replaygaining gets awfully slow with builds after 1/8).

Actually, files made with NanDub often contain garbage I've never seen DVDs containing garbage...

If your source is AVI, you DO know how long MP3 garbage lasts, because you have the avg data rate in the header.

I have, e.g. Psycho Beach Party. All audio tracks have AC3 garbage somewhere in the middle, even multiple times, and A/V sync is off after that.



I know.

Hi mosu,

Congratulations for your GREAT work

Just one question, since mkvtoolnix doesn't support it yet, is it possible to concatenate Matroska streams in Linux ?

I don't think so. At least not natively. But I think that you can run AVIMux GUI inside Wine without much hassle.

It seems to work, thanks a bunch

Note: since I don't know how drag & drop works in Wine, I can just right-click on the file listing to add files.

Thank you for an most excellent tool

@Mosu
While doom9's forum is down I'd like to report that with latest versions of mkvmerge there is a problem of 0 lenght with playing with MplayerClassic.

Regards
Lobuz

That is a known problem (and a bug) in both the Matroska DirectShow Demuxer and MPlayerClassic's own Matroska implementation. They don't handle 64bit floating point numbers at all which new mkvtoolnix builds use. Toff has modified the DShow demuxer to support those. You can download the modified demuxer at... (/me goes asking for the URL) http://www.matroska.org/~toff/test_null_sub_duration.zip Toff is pushing the changes towards Gabest (I hope ) so both MPC and the Matroska demuxer should be fixed soon.