 So, WCD. I hope everybody here knows what that means. Good, nobody's running. Right. So, what does a WCD team and software do? Anyone? It's a boff, I'm not going to talk about this stuff. Yay, give that man a prize. Yeah, basically it's the job of the software and the team to produce the official CDs and DVDs that we make in Devian. Mynd i'r cyffredin cydwyd gyda gyda'r cyfreithio dweud i gyd ddiwyddiadol i fynd i ddefnyddio'r cyffredin yng Nghymprifysgol. Ymryddon wedi rhoi, ddim bywch yn yn ôl i'r cwm director. Yn y cwm yma... Fyd pan e'rса'r mynd i amchiad gwahanol am hyn o'r cyfrifiad hynny, ddim yma i yma, i ffg, mae'r cyffredin llyfr yn ei bod yn ddondio'r cyffredin ac'r llyfr yn ddod i gyda'r cyffredin. I have put a huge amount of effort in over the years to try and make the process faster, but fundamentally it is a time consuming process just due to the sheer amount of data that gets moved and checksumed and copied and written and it's all there. We have over time started cutting down on the sets of CDs that we produce, at least in terms of the full size ISO images. We do do the work to be able to create them all, but we don't bother actually shipping all of the ISO images because fundamentally the people really need all 38 CDs for MIPSL. At the point when anybody claims that they do, I'll call them a liar. So we do try to make sure we have the first few CDs available, we have the first DVD available for every architecture, and obviously for the ones that are far and away the most common, obviously i386, AMD64 and Source, we have all of them available, and we also make the blue way images. Has anyone here actually used one of the blue way images? No, I didn't think so. They do work, honest. I've written a couple of them and they seem to work. It's a difficult thing to test, it's a difficult thing to gauge the interest in, not very many people have the drives or the media. It'll take off. Colin? Microphone. Do we actually have anyone in channel as well from outside of the room? Was the intent of the blue way images more than anything else to be able to give them lighter conferences and the like? To be honest it was a case of, I saw the drives have been available for a while and thought, sod it, I'll make one. The media for the blue way drives, at least at the point when I bought it last year, of course, was horrifically expensive. A single double layer blue way disk was, I think, about 15 quid. I haven't actually written a double layer one yet because I don't feel the need to burn that 15 quid. It's probably already the disk will have gone off and won't work anymore. I mean, it's the ones that people are likely to give away at conferences for me anyway are always going to be DVDs now. I'm not aware of anyone who really cares about CDs as opposed to DVDs. Please, if you think I'm wrong, shout, because I'm going on assumptions here that unfortunately we don't get much feedback on. So for me the main target has got to start being the DVDs quite soon. We really cannot fit even a minimal no more KDE desktop onto CD1 anymore. I know. In fact, if people actually start doing packages that are bigger than a CD, it will totally break what we do. So, in return, yes, Phil? One, two. If you're putting something on a USB stick, then a DVD might well be too big. So would it be worth doing pretend double-sized CDs that are pretty useful to USB sticks or something? No. So rather having the CD1 being bigger or something like CD1 being bigger so it's generally useful for dropping onto USB stick? Possibly. Which size of USB stick? Sorry? Which size of USB stick? Do we go for one gig, two gig, four gig, eight gig? Well, if it's eight gig then you just put DVD one on, don't you? Yeah. It's a bit of a movable feast but one gig is sort of throw away money nowadays. Two gig, four gig, eight gig, all seem fairly feasible. Obviously varies depending on how you are. Sure. I mean, actually it is a good question, thanks Phil. I mean, should we be targeting yet another different size of media in the middle and making images that will fit? So you put your bootload on the front and then we have something that will fill up the rest of your USB stick to make it two gig? For what it's worth, it's a very different kind of target but what we did in Ubuntu was we kept on generating CD images but came up with an application that turns the CD image into a USB stick and does the faffing you need to make it work properly. So rather than having to generate another set of media we just made sure that the CD target would kind of work. Now this is suboptimal in a few ways because CDs are smaller than almost any feasible USB stick these days. So we could fit more on. I mean, one thing actually, well Joey's here, you might know. At the moment of course we have the code for USB sticks that will go looking for ISO images actually on the USB file system itself and then loop back mount them and go from there. Is it feasible to actually put CD1 and CD2 in there and use them both? Okay, could it be? At the moment it just loop mounts one of them? Yes, I know that. Would it be easier to have some kind of application that emerges the two and just dumps it all straight onto the USB file system? Yes, we could. I'm just curious as to what's feasible. Yeah, as Joey says, you'd have to teach after about doing the loop mounts itself, wouldn't you? Sure, yeah. So yeah, a useful thing would be to, I mean, how complicated is the application? Presumably it's a shell script of some sort or... Wrapped up in GTK, KD. We did a Windows front end recently. I suppose it's useful. These are the people who haven't installed... Exactly, yes, I just realised that. Presumably it's all free, can we steal it? Yes. I suspect that there... I think there's a patch set out standing against Di to actually make that work, but... Fine. I mean, what's it going to say? The net install on a USB stick is exactly what we use at work for installing all our machines. Yeah. So actually that might take me on to a slightly different question. Do we still want the business card images? I ask this periodically and I hope people will eventually say no. The business card images. I find them really useful for... I find them really useful for testing Di, but I find them really useful for testing Di. I don't know if you still do, Joey. I've never really found them that useful for testing Di personally. The business card images. I mean, there is media, but I really don't see people using it much to you. Yeah, no, not at all. I mean, it doesn't hurt us a huge amount. It maybe adds about five minutes in total to the build. It's just the extra complexity, it's an extra set of images to watch. And they tend to break more easily than the net is. Absolutely, yeah, they're very fragile because of course the moment that anything changes in the archive it's all broken. So... No one said... Well, Colin said yes, he likes them. Does anyone who counters like them? I can build that sort of shit myself if I have to. I don't use the actual generated ones on CD image. Yeah. I mean, are there any other images that people think it will be useful to have? I mean, I've been pondwing for a while, as I said, we can't fit everything onto CD1 anymore. I don't think... I'm not convinced it's actually worth the effort of trying. What do people think? Are they called CD1 for no one KDE actually very useful? Or should we possibly admit for the first time that CD1 on its own isn't enough? I actually believe that maybe a very small amount of extra application could be useful on a desktop. I've had a look at some of the distribution and some other distribution ships extra packages than we do. It has to be discussed with us sensible, but to answer your question, I think CD1, yeah, who still uses CD and maybe having a system so you can just have a CD1 which is bringing a minimalist x desktop and have something to let people say, you didn't have network, so you start with your tiny X or X session and you have to fix your network connectivity to be able to get the remaining stuff, but really CD1 is so small. Okay, I guess my experience with CD1 is that if I'm actually doing an installation somewhere where there's not good network, I make sure that I bring the DVD to go off and hit the network, and I know that I can avoid hitting the network but then I'm going to get some minimized desktop and I'm not quite sure of the state of it, so I think it would be a lot easier on us on the test all side if we didn't have to really worry about just size quite as much as we did. Yeah. Fans and I put a lot of effort in at the end of last year to make the first DVD more useful, so it will cover all of the tasks for known KDE with as many other languages as we could make fit. Even that was hard. By the time we want to do that for i386 and AMD64. I mean, I do still like the whole multi-arch desk idea. I hope other people agree with me. If not, fine, we'll drop it. The multi-arch desk seems very useful for giving away at events. I'd be surprised if you often find people in a position where they want to use more than one of it and more than one of the architecture is available on it. Yes. To be honest, it's probably becoming moot in the fact that everything is AMD64 now. Any new machines? Hardware does support it. Yes, of course. The Atoms are still i386. Until we actually do multi-arch, the DeepCup side, AMD64 is sufficiently painful in user space that a lot of people avoid it. Actually, having said that everything is AMD64, yes, again, all the machines we install at work are brand new quad core boxes. All the tool chains we get from the vendors which is what the boxes are for are still i386. Right, of course they are. I do wish they'd fixed that. So, okay. Yes, the multi-arch DVD is good. Per, do you have something? Yes, we have a question from IRC. CD1 base plus sum and CD2, KDE and CD3 GNOME. Would that fit? Maybe. One of the funnest bits about trying to do clever arrangements of the bits that go on each of the CDs is I have no idea what belongs in KDE and GNOME. The task cell people have some idea but there's a big flame war periodically because their idea and say the GNOME team's idea don't quite match and so the white applications don't get on the CD. If we can come up with a good way of recognising things reliably, programmatically, without me having to worry about did GNOME make it onto CD1, for example. I'm all in favour. If someone can provide patches or give me a good way, a good description of what exactly to do there, then yes, great. Please, we can go with it. Quick show of hands in the audience. How many people here would actually use ACD? Joey's already said that he basically wouldn't on his own. Okay, fill up the back will. We do have a few. I say use ACD, I mean as in install the machine just using CD1. Yes, fine. Okay, a normal full CD image. I quite often use CD1. I would very rarely give a damn about any of the desktop stuff. So having all the things for doing a server install on CD1 would be really useful. And it means that if for some reason the local network is screwed or the hardware doesn't have drivers until you do something else, then you can get something pretty useful onto a machine from CD1. Would it be worthwhile shuffling things around in that case make CD1 be not just the base system, but include on CD1 whatever you need to actually get the rest of the system up. You know, open SSH server, maybe even compiler, or is that too much? It probably is. I can't tell you off the top of my head at the moment. If it isn't then CD1 is pretty useless to me. Yeah, sure. I mean should we actually do that and give up on trying to fit any of the full desktops onto CD1 and just say fine you'll need two and three probably. I have an indirect question. Currently aptitudes are installed and recommended by default. Recommended dependencies. And the CD doesn't seem to have the same behaviour. Do we... So it seems to have the same behaviour? It doesn't behave the same way. When you install it from CD you don't have the recommended dependencies whereas you would have them if you just aptitude or synaptics. Is there some... That's only just changed quite recently. Also deliberately the WNCD code itself when laying out the CDs and DVDs does not look at recommends. For what it's worth I think that's a mistake once apt is installing recommends itself. Yes, absolutely. Because then you'll be missing things. Well it's also very difficult to tell apt to go back and reprocess recommends letter. So I don't think that's something we want users to have to do. So at some point in that case basically when we're laying out the CDs I mean hands up to anyone here who knows how the CD code works that lays things out. Yeah. Basically it goes through we sort the list of we work out the list of the packages that we want at the very first things in there are things that are required and are listed in WNCD itself. So we know that we must have a couple of packages at the front just because it breaks dependencies otherwise. Then we go through whatever tasks that you have defined and so we pull in a basic set and that can obviously for the official CDs is a fairly obvious thing straight from Task Cell. It can also if you're running WNCD locally include whatever the hell else you like and then if you do a full complete build it will then pull in all of the information from all the packages files and will sort all of the packages it knows about so the dependencies are met as well as possible. So you should never end up with the dependencies for something on CD4 or actually on CD7 it's sorted appropriately. The moment we start having to take recommendations into account the only sensible thing we can do is treat them as dependency in terms of how we lay out the CDs. I have no idea how much that's going to break things. We certainly found in Ubuntu it blue things up a reasonable amount it wasn't it wasn't completely unreasonable but it did require some. I'm still very happy it must be said how well we do cope with the dependency sorting and the fact that aptitude whatever do the right thing in terms of they know which packages you want and where they are so you only have to feed each disk once. I was horrified to play with a Fedora install quite recently while watching somebody else at work do it and it's shuffling disks like crazy. Why? It's not hard. Well I don't think it's hard we manage it so it can't be that hard. At some point again I guess we will need to start thinking about recommendations as though that depends that is probably going to utterly destroy any chance of putting a useful graphical desktop on CD1. Just a heads up I hope nobody here is going to disagree with that. Are you going to argue? Peta? Yes Yes Yes Absolutely Yes Yes Yes Yes Yes Yes I agree, absolutely. Again I hope somebody is taking notes about all this because I'll have forgotten who you are and what we spoke about in about 20 minutes. So I recommend going in people still like CD1 but I honestly don't think it's going to work well for desktop for much longer. Is it worth still trying to make it that happen or do we admit failure or is that something we should do the numbers for once we want to remove to recommend? Possibly I mean, it would possibly be worth going with a lightweight desktop, yes. We already have the third option for CD number one which is to have the LXDE slash XFCE and that works quite well. You have a lot more other packages on the CD because you don't have so much I hesitate to call it crap but you don't have quite so much other stuff just needed for the desktop. That might work for initial CD1. We'll have a look. If we move to recommend, is that going to take more or less than four months to fix? That's a very good question. I can turn it on tomorrow and wait for the breakage. Fair enough. Another point about the recommend is that having them on the CD will make the CD behave more like the network installation because all the packages are on the network. Yes. What else was I going to talk about? Does anybody else have anything to bring up while I think about that? Are there any particular modifications to what we do that people would like to see? Maintaining simple CD which is largely a wrapper around the CD that generates a customized CD the way it's handled during release cycles is kind of terrifying. Three days before the final freeze there's a new Debian CD out and I'm like well I can test it with Subversion but if there's one change that I missed Apologies for that. It has always been the way it probably will continue to be the way. I don't like making changes just before release but Sodslaw says that there will be something changes in the archive that we've got to cope with in the last week. Debian CD is typically one of the very last packages to be installed in the archive just before the release. Actually if you haven't seen how this works I can tell you that there are possibilities in the sections of the archive even though we told people there are no uploads anymore so we can just hand push packages and I think that has happened to the already more than once. And obviously it's necessary so for my lease when I was a lease manager I said well it's not a little risk for us to do it because we know that the packages have been properly tested, we have some things and if it breaks well it's deeper out anyway so yeah. Most of the Debian CD changes are most of the very let ones are task changes aren't they? That kind of thing yes. Would it be worth considering moving that out of the source code and making it a property set in the archive exported in the packages files rather than done in tasks just to try to reduce the amount of changes. Well in fact as time has gone on we actually again one of the things that Franz and I put a lot of effort in just before the end of last year was we finally got away from we tried to get away from using the task cell stuff shipped in the package so we generate as much as we can automatically of course we do want to make sure you can't win here we then also wanted to make sure that the stuff that was in the package matched closely what actually shipped in Lenny so people without net access without everything could generate similar looking CDs afterwards. Sorry I had a separate question about the publishing that we do we do things like bit torrent tracking CDs don't we? How do we find the tracker software that's used? Oh it's all crap. We've found that it's all dreadful Yes it really is. Is there anybody working on anything that is not permanently sucky? Yeah we found especially doing the Lenny release that we got a whole load of friendly people all over the world with nice fast connections and we told them yes please help seed. The tracker fell over. Then the tracker fell over or all these people yes we're running cedars and the cedars just would randomly get bored and stop doing anything and nobody understands why and it's just crap. Is this worth a summer of good idea for next year? Oh god yes. Right one that is not dreadful. Or if anyone knows of anyone who is working on a good torrent infrastructure code please God encourage them. We deliberately generate as part of the weekly cd build and also obviously of the official beaters and releases we deliberately make sure now that we generate a torrent to go with everything we generate zed sync files scattered all over basically whatever we can do to help people download stuff without hammering or bandwidth totally is good. If there's anything else that we could do there I'm willing to listen. I know I have had metalink folks talk to me several times do you have any experience of that? We had people wanting metalink for Windows best downloaders apparently it works better. We did start generating them we avoided most of the really scary features of metalink it has several kind of server hammering features like opening multiple connections of all sorts of things. So we arranged for that I think you can turn most of that off in a metalink file we just gave it a really simple file that's essentially just a wrapper on bit torrent that seems to work okay from what I understand but it is a bit of a pen that could just be using the torrents. Sure. Do you have a question? What do you think about adding the translation file to the CD? I understand that there are some size issues for the CD but there are also other media like DVD just for information the whole BZ2 file that's 7 megabyte currently per language. No no no the full list is currently 7 languages and the German one is 1.2 okay cool and there are 30 languages. Sure I remember you asking this before we did Lenny and saying yes I'll have a look and I failed I'm sorry I was crap. There are also some tiny issues there. I would really like to put the translated packages package data on to each of the CDs and DVDs as well. It's a little bit more complicated than just dropping the existing translated file. Again I'll describe a little bit why the way that we build up a CD these days is we actually generate a temporary ISO image just to size things. We then go through and iterate through adding packages guessing how big it's going to be and then periodically actually generating again the ISO image just to test the size keep going keep going keep going so when we add a new package we have to make sure that it's available for the architectures in the white order and Ditto source is considered an extra architecture there but also this is there probably more detail than you need the other thing that goes on along with the package is it's entry in the packages file and then of course we have to recompress the packages file to make the packages.gz the packages.bz2 don't know if we'll actually do the bz2 or not if we want to do the translations to match of course we will then have to go through and add the equivalent entry in each of the language versions when the way that we know that a CD is full is that suddenly is that we had a package that doesn't fit anymore and so we actually have code and I'm really it's horrible nasty code I wish somebody could would volunteer to come and have a look but my pearl isn't very good once we actually overflow and know that a CD is too full we actually go back and undo it all we remove the package we remove the packages entry so we'd have to do the same thing in all of the different languages as well that was why I said yes we'll do it and then went quiet it may sound like a horrific way of doing things but what we used to have was a real problem where we would get up front what would fit on a CD we'd get up front what would fit on a CD we'd get right to the end of the process and say oh crap it doesn't we used to just have hard coded fudge factors we did and it was always total guesswork we'd end up with a CD that might be a megabyte bigger than what we'd be happy with do we don't we because of course if you're doing a very large set of CD images you only realise at the end of the run that CD17 doesn't look the right size if you want to redo things you then end up got to tweak a fudge factor or tweak your maximum CD size and rebuild the entire set that just does not work at all anymore because going back to the woody days for example we ended up on some of the point releases spending an entire week tweaking the CD sizes to make sure that none of them was too big but to try and avoid having the 35 megabyte image on the end that just would be a total pain so one of the reasons why I moved over to the sizing by doing genizer image for WNCD for version 3 was to move away from that we can now know exactly how big a blank CD is exactly how big a blank DVD is and we can test the sizes of what we've built against those and we know exactly is it going to work or not if you look at the sizes of the typical CDs on CD image you'll see that most of them are within a megabyte full obviously that means that over 35 CDs the chances are we've just saved somebody a blank we've saved somebody from having to carry out around an extra CD and of course the more packages we've got the more important it is to pack them in tightly yeah it's hard if somebody wants to repair patches for the translated package descriptions and stuff that would be wonderful I think we already have a script which can receive a package files and provide a translation file for matching this packages file okay that would be cool something I can attach to Berg cool now again for performance reasons we don't go looking up the details of outage package as we're running we actually in WNCD in the middle of the dirty great lump of pearl that lays out the images we suck in the entirety of the packages file up front so all we're doing is just writing out data that we already have in memory we will end up having to basically have an array of those one for the main one and I guess another entry for each of the translated descriptions maybe ah am I dying right can you hear me again yeah sounds good um what else yes um jig do stuff do people still find it useful I hope so yeah good okay I mean have you guys actually moved over to the new WNCD yet or are you still doing really really old stuff we unfortunately haven't and this is because I haven't figured out how to cope with the subversion to get switch for our vision control yet um we will get there subversion to get ages ago we're on the last code just before it moves to subversion we're not in yet oh okay fine just but I'd love to move to get pick your pair of vision control systems fine okay so we have to cope with that I'd really love to but it's been somewhere in the pile yeah I know it doesn't bite us all that badly because we are stressing the we don't have these you know 20 CD sets and so we can essentially afford to deal with it it is a bit of a pin but but it's tolerable so um sizing is going to become even more important if we're going to go for recommend if we're going to go for translated package descriptions um we're about to kill cd1 I don't think it's going to work for much longer so at that point actually coming up maybe even with a hand picked set of packages that we really want on cd1 and then we lay out beyond that it's probably going to have to happen again so if we can get the di folks to allow us to have more than one eyes of image on a USB stick or if we can go for the Ubuntu style you know we lay things out on the first couple of images I mean back in the dim and distant past when I first started on the Debian CD 3 stuff you might remember this was back in like Helsinki my plan was very much that we would include in the Debian CD package a tool to allow you to pick out an existing image that's already downloaded and basically roll back the last few packages that went on so you could make yourself extra space to put on whatever else you wanted I haven't got that I don't know if I ever will well all of the information that you need to be able to do that should be on the CD it's just a case of doing it if anybody is really looking for something to dive in but I don't know would this be useful for the CDD stuff or not talking about freeing up a little space at the end of the disk to put other junk in yes most of the stuff that simple CDD uses is already in packages sure oh absolutely there's like a few scripts or something oh yeah, no what I'm saying is it's not necessarily just adding whatever random files onto the CDD you want but it could be specifically we've gone through and filled up the entire CD image with packages but you may want to roll back the last 20 to add on specific packages say for Debbie and Edu or for whoever else has their own set of packages I mean I don't know if it's useful or not if it's just pying the sky idea but it does mean you could start off with an official CD and then just tweak the very last few things on it or do you believe it will not be useful because we need to change completely the list of packages on the CD fine, okay in that case I'll stop beating myself up about it the jiddo stuff people say is still useful one of the things that I'm going to be doing when I was going to do this week yeah like I was going to get the time is moving or at least porting over the jiddo generation code that is currently in CDR kit into oh the little not libcdio the libburn yes there is an ISO image generating program to go with that and really onewing off topic slightly speaking as one of the upstreams of CDR kit I would like to see it go away and actually throw away all of the shilly related code we have in the archive how well does libburn work for these niche stuff like HFS hybrids I knew you were going to ask that at the moment they just don't it's a hard problem I mean to be honest isn't it effectively a mess yeah it is are people aware of do people know what we mean by HFS hybrid I'll explain briefly basically if you want your CD to boot on a mac if you want all the long file names and various other bits and pieces to work on a mac you can't just use ISO9660 like you can for the rest of the world you've also got to put a whole extra parallel set of file system descriptors in the disk that make it look a bit like HFS it's not quite HFS really but if you squint and look a scant at it yes it comes out looking much the same so Macs will deal with it I wish they just give up and do something portable I don't know if so that was for the ppc mac world I don't know if this has changed since last time I looked at Intel Macs it's may have done but when I last looked at Intel Macs you still couldn't boot straight off ISO9660 you had to have HFS plus hybrid and nothing that I know of does I don't even know of the software to generate a bootable HFS plus image at all let alone a hybrid please God well somebody's going to slap them till they stop doing this shit if you crawl dfi in weird way something you can eventually get a boot off ISO but it doesn't really like it so basically the problem is this was first a problem for us with m68k we had to do all kinds of horrible and mobile things to the image to make it boot on m68k macs, amigas and ataris to be honest the 68k had gone away as far as we're concerned we stopped making images for it a while ago but of course as Apple moved on to power pc this means that every image that we created for power pc has HFS hybrid options turned on A that means that it slows down the generation code a lot the source code for doing the HFS hybrid inside mkfs both the old version that we've ported from and still in current versions is utterly shit it really is when you start generating large images when we first started doing dvd images I called so many crashes in that code it was untrue the reason that we haven't generated any blueway size images to go on power pc is because that code just fails isn't that just I seem to remember there's a knob that says you need to make the HFS file descriptor table this big in the end it came down to that yeah but if you actually have to look at the code inside though that does it but four years ago it ends up with cursing it goes through as you generate your HFS hybrid descriptors the size of various lumps inside the HFS volume area I think I forget the exact terminology changes so as you get a larger and larger image and you start growing things you can end up spending whole minutes inside genizer image just calculating the sizes of that stuff so the power pc dvds take about four times as long to make as the i386 ones do for exactly this reason it's nasty so I would love to just give up on it and I was hoping that Apple would have seen sense with the new intel based stuff but apparently not but it's not like there's anything we can do with it like I say I last looked at this relatively it was a year or so after intel max came on the scene and it was still very early days you had to they didn't ship with bootcamp but default yet at that point so it is possible that the universe has become a little more since then fine sure I mean so along with the jig do patches that went into macaws of fs at the time there's also a whole load of stuff in there to support booting from several of the other architectures I know Steve Langershire got involved with that even more so which is how we managed to end up with the hybrid HPPA IA64 alpha CD they actually all need pointers in the first boot block pointing to various files on the disk but we actually worked out a way where we could interleave that those pointers without confusing anyone and oh it's horrible but it does work but as macaws of ffs and now gen iso image frankly are going away as far as I'm concerned they're legacy only somebody somewhere is going to have to forward port a whole load of that code into and I'm going it's really crap I can't remember the name of the Libburn attached iso image tool nor can I sadly there have been bugs filed against Libburn asking for HFS hybrid support for about three years I think so we probably just need to somebody needs to do it pay the Libburn upstream or something like that so I can't really imagine anyone else is going to do the porting of this, of all of the jigg do kerfn whatever so that was my work plan for this week and then people got me to do all the talks and it's all your fault so that I'm hoping might happen before squeeze but it does need me to spend a couple of weeks on porting over that code and I'm hoping that the Libburn back end is reasonably portable and is nice and easy to follow please any other questions any other comments, any other ideas really? Regarding the bootloader for E386 or AMV64 just any about having language multi-language support or improvement, maybe something about being able to display hyper-screen on the front page are there any progress on my stream to be able to help with that and that's what we were talking on Dash Boot recently about doing a Facebook front screen since that's now actually supported by Cislinux or by Cislinux upstream so that would give us that kind of ability I'm not quite sure where it sits between the installer and WNCD it's somewhere in the grotty intersection the way that we currently handle the configuration files for Cislinux is a bit scary let's be honest the DI folks have a whole load of different snippets of config for working with the graphical menus and stuff that inside WNCD we then do a whole massive set of horrible munging with friends again worked on this when I was with him at the tail end of last year so this is how we can generate a set of menus that will match what's on the CD whether it's a no more KDE or a combination if we want to start doing that with the graphical image as well I don't know I don't know that it makes very much difference what the front end looks like there is some horrible does Cislinux end up doing font rendering and stuff then what you can do is you can get the graphical front end thing to read through the Cislinux menu files and do the right thing there's some it's not quite as easy as all that unfortunately but it's not necessarily utter doom and having a CD it's only mild doom that's fine we can deal with that yes we have a few sorry we have a few questions on IRC if I can find them you can call out names as well to go whether that will be appreciated MRVN will CDDVD1 have some space for firmware or will there be a CDDVD image with non free that's a very good question one of the things I was hoping we would have out of discussion after the Lenny release my own preferred way for things to go would be that we would have a firmware section that we could then agree as a project would fit on the first CD first DVD and then that would make the question moot we haven't had that I don't know if it's going to happen so what we have at the moment is probably going to carry on for the foreseeable Joey put work into DI I put a little bit of tiny bit of work into W and CD to match up and that is the extra small downloadable lump of firmware that we currently have until we end up with a better solution for the firmware and we're done with the same route ready for the next one? PCC asks what compression method is currently used would it help to use something like LZMA compression for what? good question if you can clarify the question that would help we can take another one in the meantime the mill asks if all the IA64 images could print a message saying use the AMD64 image you idiot yes that would be it's something that people have suggested before I don't know how to do it in theory it should be feasible to put a normal I386 style El Torito boot image on the front of the IA64 so if you attempt to use it so if you attempt to use it on a non-IA64 machine then it will print it will just come up with it will run something that says you've got the wrong CD yes that should be doable patch is welcome I have no idea Colin just one last clarification well Colin respond it would probably also help to look through the website side of things and improve things there since by the time somebody's already downloaded an IA64 CD it would probably already print I mean we do the two most common questions we get in Debian CD or in Debian about the images are help why is this DVD only 400 megabytes because people have downloaded it using quap software they're using fat32 and it doesn't work and the second one is I've downloaded a 64 bit image why doesn't it boot exactly as the question as you just forwarded so yes we should really do that too there was a further question which of course I've further option there which I've just totally and utterly forgotten somebody else had come up with some extra ideas on clever stuff we could do with El Torrito boot it'll come back to me probably just after we've finished okay next question just to clarify the compression question was about the devs oh the devs fine, at some point we already have support I believe for using ElzaDemma and whatever yes of course as the devs themselves get smaller we'll get better but more stuff on the CDs yes we like that it's easy to do that selectively I'd caution against trying to do it across the board in many cases it doesn't actually save in small cases or particular classes of files that doesn't save much and you do have to be quite careful about memory consumption with ElzaDemma I'm told that we're out of time I will cheat and say one more point better regarding compressing I think the time is better spent fixing the compiler to make all the same size binaries it used to do in Woody or binaries has been blowing up in size with no obvious improvement so right if there are any more questions obviously I'll be around on hash debbyn cd or on the debbyn cd mailing list if anybody did get notes of all of this please forward them on otherwise somebody's going to have to watch the video and I hate looking at myself on video okay cool thank you very much bye