20120902-162635.jpg

Sunday

OpenELEC

At this point, OpenELEC is very well branded and somewhat unsynced from the XBMC dev cycle. There’s a question about whether we should push to pull OpenELEC fully into XBMC and call it XBMC powered by OpenELEC or OpenELEC powered by XBMC, or something.

At the moment, an idea is that OpenELEC could act as a sort of XBMC Stable or LTS branch, where both XBMC and the underlying OS are rock solid. Fully integrating OpenELEC and standard XBMC is looking like a long term project that may never actually happen, in the sense that we may always do something like releasing Frodo, and then intentionally hold OpenELEC back until 3 or 4 months later to ensure stability.

What ends up happening remains up in the air.

New Project Managers

Keith and Jim have agreed to become additional individuals who will be available to answer contact emails and work with businesses interested in working with XBMC, in addition to Cory, Pike, and myself.

TheTVDb

The TVDB recently lost much of its free internet hosting. Now, they need ~$500/month. The TVDB is massively critical for XBMC, so we need to review how we can help. One suggestion is to overhaul our UI so that when users add shows or movies or whatever, the scanner actually displays our scraper site (e.g. TVDB, MovieDB, FanArt.tv, etc.). In theory, the more other people actually SEE where their info is coming from, the more likely they are to get better donations.

Alternatively, any sites that we absolutely depend on we could work on trying to offload some server time. This is less easy to accomplish, but at least worth looking into.

Meanwhile, I need to do a post about all of the sites that we absolutely depend on a daily basis.

Saturday

Internal Discussion

4:30PM Vienna Time

Frodo

Current code stoppers: Code generation, PVR, Raspberry Pi are the three big pending changes.

Arne (aka Spiff) will be acting as the Frodo release manager. This is a newly created position that essentially says, “Beta 1 happens on this day. We are now ready for RC1. Time to release.”

Other discussions during this period include how to better manage the way XBMC deals with addons during the Beta period and how to handle skins that are missing certain xml pages and links. The typical question is, “Do we let a skin into Frodo, if it doesn’t support PVR?”

PVR

Lars is now Skyping in to talk about PVR. Right now the primary issue is that we need more people to work on PVR. In the past, the largest problem has been binary plugins. The first suggestion is simply to drop binary plugins. That’s possible now.

So… there’s really no other stopper. Without binaries, PVR is possible now. The September Cycle is going to be an exciting one!

One important issue is that PVR is the first bit of XBMC code where we don’t provide all the software to make things happen. That means that those users not directly plugged in to the XBMC blog/forum/wiki are going to be very confused when they click the PVR button and nothing actually happens, because they don’t have a TV tuner or TV decoder software. The currently accepted solution to this issue appears to be to create a pop up that says, “You don’t have a TV tuner and TV decoder software connected, and you need them. Please visit xbmc.org/PVR (or something) to learn more.”

3:00PM Vienna Time

Monthly Merge Window

We are now reviewing the monthly merge window. As you recall, each month we have 10 days to merge, 20 days to clean code. There is a debate whether to increase the merge window, whether to set aside a portion of the merge window for big updates, or what.

The primary issue is that code change is difficult to manage. While Mozilla has the resources to employ 20 or 40 builders and testers, 30ish people is our entire dev team.

One solution to breaking is to try to break up big commits into small packages that each get pushed in, so we can know which small portions appear to be breaking and which aren’t.

Trac

Possibly somebody else can fill this in. I pretty much blanked out for this entire section. Mostly, it was Github = not everything we want. We have a ton of open bugs that are probably super old and out of date. Many don’t actually look at Trac.

Documentation

Ned Scott explains mediawiki! In particular, we learn about python scripts, and templates.

12:30PM

GSOC

Jonathan summarized GSoC by saying that GSoC is awesome. We had four projects that we semi-called for at the beginning of this year’s project and they’ve done a great job greatly improving our library. It’d be nice to have more mentors this year and also a GitHub admin whose whole job was to harass students to make sure they were doing weekly code.

We’d also like to better work with the team calendar so we have a better chance of making sure we actually get into GSoC every year.

Amejia’s project comes next. The goal for Amejia’s project was to create a series of tests to make sure various aspects of XBMC work underneath the hood without needing error reports. For example, archiving within C++ does not appear to have been working for the past 10 years, ever since it was first written into XBMC.

Montellese came first. The situation that his GSoC was intended to deal with was users with huge video and music libraries and very little way to to easily get to each item.

His solution was to use Smart Playlists (essentially smart filters to narrow video and music lists). Unfortunately, Smart Playlists have not been very easy to use.

His plan was to speed up these SPs, integrate them into the library, etc.

First, he had to refactor sorting and limiting functionality. Then he added the ability to short or limit by artist, in addition to albums.

Then, he wanted to extend rule searches. For example, after his work, you can do a search for a list of movies of Nic Cage and Bruce Willis.

This is only available in Library mode, as it depends on the database functionality in the Library. Filters differ based on type. So you could sort by type of instrument in music, but you wouldn’t get to sort by type of instrument in movies, because that wouldn’t make sense.

You can actually see Filter in current nightlies.

Topfs2 (aka Tobias) came second. His GSoC was a Clean Scraping API. Tobias has put together an information extraction program called Heimdall with the goal of not worrying about content type (movies/music/tv/sports/whatever). Additionally, both outside scrapers and XBMC could potentially use it together. Finally, scrapers typically work right now by first looking for a title, then looking for built-in metadata, then reading the stream, then finally visiting a scraper website such as TheMovieDb. With Heimdall, we can perform the scrape on the computer separately from scraping on the website, so scrapes can perform faster and more efficiently.

There are no plans to fully integrate or replace any current XBMC scrapers at this point, but future potential is high.

Alcoheca has updated the latest UPnP server, started cleaning up code, etc. He’s now added a port for media control, so if you have 2 XBMC devices in your home, you can watch a percentage of the show on one device, then send the rest of that show to another device. Additionally, the XBMC device essentially acts as a remote control for another XBMC instance. In essence, this works somewhat like AirPlay, but with more extensive possibilities. Additionally, the uPnP client is being extended to better show metadata.

This isn’t available yet, but it is on the horizon in the near future.

Company Presentations

11:15AM (Vienna Time)

AT-Visions (the primary host of this year’s DevCon. Thanks guys!): They use XBMC as their frontend for Hotel IPTV systems. It actually looks quite a bit like XBMC, except with extended functions that work more with Hotel systems.

They’ve moved from NVIDIA ION to AMD Fusion chipset (Linux-based). Install base is approximately 150,000 machines. They are looking into ARM boxes to reduce cost, but so far they really don’t seem to be up to speed for business services. Their primary ARM testbed so far has been the Raspberry Pi, but they’re interested in taking look at the Pivos box.

Technical issues with the current version were discussed from video playback to mysql database problems to multiple different fonts running on the screen at the same time. Occasionally there were between 3 and 5 conversations happening at once. It was exactly like watching 5 different IRC chats happening at the same time!

10:45 (Vienna Time)

JetStream: A company who sells media servers to very high class yachts has incorporated XBMC into their service. JetStream uses ViaSat to stream video content over satellite to boats. While it’s a pretty great service, the latency is fairly high (as is only natural for satellite)… so probably no gaming.

JetStream acts as a placeshifting service, essentially like Slingbox, except instead of placeshifting to your work, you placeshift to your yacht anywhere around the world. Everything works simply and in full time. To be legal, everything requires access controls. A yacht owner buys a satellite (or, typically, numerous satellites throughout the world) which is then placed in the appropriate region of the world. For example the owner actually buys a lot in the UK for SkyTV. Then JetStream takes in that content and sends it directly to the owner.

For those of you interested, while the presenter didn’t give us a ballpark price, he said he himself could not afford it. Essentially, this is a service ONLY for those who can also afford yachts that cost more than the respective homes of possibly every single XBMC dev combined.

10:00 (Vienna Time)

Seagate The Seagate presentation is still very much in early stages. For now, they’re mostly interested in potentially using XBMC as a headless uPnP solution or something similar. More conversation is definitely necessary.

9:00 (Vienna Time)

Sigma Designs: As you may be aware, Sigma has been working with a few developers to make XBMC available on their SMP8656 (available now) & SMP8672 (port in progress, being demoed at Computex 2012) chipsets. Because these chipset CPUs are so limited, Sigma had done extensive work to totally separate the video decoding process from the CPU, so now any video playback XBMC lose hardly any CPU cycles. To put it simply, XBMC’s UI running on top of video playback works pretty well. Adddons (particularly scripts like PseudoTV) tend to be fairly slow, but to some extent that’s because Python in XBMC is always fairly slow. Plugin Addons (i.e. addons that let you watch online video and listen to online music) work well.

Where these chipsets get really exciting is in two categories. First, they’re MUCH cheaper than x86 processors and easy to implement in just about any system. Second, because the player is totally separate from XBMC, the Sigma player can do fully licensed DVD and Bluray navigation.

8:45AM (Vienna Time)

Nathan (me) sent around a form for XBMC Foundation members to affirm the Foundation By-Laws. This is the last step in joining the Foundation.

Ned Scott is doing his best to record a video of all the proceedings. With luck, we might be able to upload that somewhere for everyone to see.

3:30AM (Vienna Time)

You may notice that we don’t have any live blog info from Friday. I’ll try to go back and fill in, but long story short, Ned Scott, Cory, and I were stuck in the Frankfurt Airport for 8 hours right in the middle of a labor dispute, and after getting on stand-by with a guarantee of definitely not getting to DevCon on time, we tried to get on a train (none were hitting Vienna until an hour from now today) and finally, we managed to rent a car and drive the whole way. So much fun!

Thursday

8:30AM (Central Standard Time, Kansas, U.S.)

To avoid excessively clogging the various XBMC social networks with news about this year’s devcon, I plan on exclusively liveblogging what’s going on using this post. There may not be many updates today or tomorrow, but I’ll do my best to cover as much of Saturday and Sunday as I can. So check back here on those days for more interesting material on this post.

Right now, we’re working out the details for how we’re all going to meet up tomorrow and various other last minute prep items.

Those of you interested in the meetup, I’m still trying to work out details. One user has suggested the Schweitzer Haus, and I’m trying to get one of our local people on the case of either confirming that location or resolving on something slightly closer to the conference center.

Team member keith has just arrived in Frankfurt, a hub many of us are passing through, and has informed us that Frankfurt has no free wifi.

10:30AM (Central Standard Time, Kansas, U.S.)

On the plane bound for Atlanta, where I will be meeting Cory (aka TheUni). Only like 15 more hours!

2:16PM (Eastern Standard Time, Atlanta, U.S.)

I’ve arrived in Atlanta. Unfortunately, there’s still no update on the Meetup. Flight leaves in approximately 4 hours. Here begins THE GREAT QUEST TO ENTERTAIN MYSELF!!!

Somewhere over the Atlantic.

Midway through the flight, the video system reset itself, and Cory and I discovered that the system was based on a 2002 version of WindowsCE. It’s really remarkable how much better XBMC is than a custom embedded Windows build from a decade ago. Anybody know any higher up from Airbus or Lufthansa? We’ve got a suggestion for them… ;)

20120831-001832.jpg


  • M_Auer1

    Hey Nate,
    on the ways back you’re again in Frankfurt?
    We might have a beer on me there ;)
    cheers,
    mad-max

  • Pingback: XBMC DevCon 2012 | Tommy Winther

  • http://profiles.google.com/pseudo7 Chris Lee

    So who’s the guy in the pic – NedScott? (the awesome legend of the wiki?)

  • http://profiles.google.com/pseudo7 Chris Lee

    Also you should tweet/facebook again just to let people know your live blogging has started – interesting stuff being thrown up :)

  • Savell

    OpenELEC within XBMC working together would be the best thing since sliced bread!
    I absolutely love the idea of OpenELEC (Currently run it myself, making my own Frodo builds already) but to have it working with the XBMC devs can only bring it to new greatness!

  • Harley Davidson

    Any blog info from Friday yet?