 So, first of all, welcome to this boss we originally had as scheduled as talk. We now converted a bit more to a boss because there are a lot of things we should discuss about. This is a stable release team, and also these people who are connected, who are working quite much for it, even if they are not an official member of the team. So first of the introductions, my name is Andy Bart. I'm one of the two stable release managers. The first stable release team are Dan, a Luke, who is also of some interest here. So, of course, France was the installer, HHS as the keymaster. Steve is the former Anstelle release manager. Release wizard. I'm not saying the other titles are not the slides. So, we have a lot of people who know Sam, Voucher. So, okay, I feel like there are a few people that don't know Sam. Okay, we are... This is the first Debian conference for us. Okay. And, okay, we are working in Nokia and doing this malware stuff. And, okay, I was thinking that... I'm not totally sure if this is the right place to be, but I was thinking that if there is something related to release management and so on. So, yeah. The first question is that basically what you are going to discuss and... Basically, my plan for now was to just say a few words. How do stable releases work? Because not everybody knows it. And then we can discuss topics that come up from that that people already have brought with them. I hope that sounds like a good plan. Yeah, okay. Yeah, this first part, how it works and so on. So, to get to understanding, yeah. I'm sorry, did you say what your names were? Jannemanninen. One of those, sir. One of those you could also pronounce if you wish. So, we should be making up nicknames now. That would work. Jannemanninen. Yeah. And I used to want to... I used to be used to... I almost never learned the show now. I can read the letters. You must... Well, pronounce. Okay, anyway, so now... Basically, intrestable short? Yeah. That's okay, of course. That's one of these value such things. Thank you. Okay. Yeah, yeah. So, I'm with Jannem and I'm also with Yiponokia. Same for the magazine, same for the general magazine. So, yeah, just feel free to talk. So, you're making up? I'm Frankie. I was involved a bit with All-O-Tile, if you remember. Yes. The web part of All-O-Tile. Basically, I'm interested in listening to what's the changes that they will release later. So, something is changing again. Okay. So, then I will just tell you how currently stable releases work. What we have, or what you want to know is that it's basically we have all the packages in one suite in the FTP master called Stable that's a package of stay. Then if somebody has a change to upload, we should, of course, first speak with a stable release team because the changes looks okay for us. Then if the change is uploaded, it goes to a separate directory of the FTP master. Yeah, we could then use a package and say, yes, it's okay, or no, it's not okay. Or just leave it there and send it there. At least another drive would be nice. And if you accept the package, it's now automatically installed in the so-called suite proposed updates. And on the points release, the packages from the proposed updates are then installed in the stable suite used to be such generated. So, it doesn't look too complicated. There are a lot of things involved. For example, if we update the new kernel, things probably need to change for the installer. So, the new installer needs to be uploaded, which was just done. Then the CD which just needs to be released and that was the technical side of how stable updates work. There's a non-technical side, so what do we want to have in a stable update? What do we accept as packages? For us, it's very important to say that things shouldn't break for the users of deviant. I think that's our uttermost goal to say it needs to continue working. And so, for some times, there are changes which are very important. For example, if we have a security change, we might need to accept things that are not so nice. I can remember there's a Zulu update, so that was a bad example. Yeah, for some bar. That's a more recent example. That was a mistake. That was a bug at least. That was a mistake. Of course, mistakes happen. That's one thing, but the other is with Zulu, we are going to break configurations, but we consider it to be more safe to break stuff, which doesn't make me happy, but I wasn't in the role to make any decisions at such a time anyways. So, I can just say it sometimes happens, but usually it doesn't. What we are going now as well is to say, sometimes there are really very disturbing bugs, where fixing stuff isn't risky at all, but would have already been so. For example, if we add new PCIIDs to an existing piece of hardware, that might have users, but it's not really risky. So, we can still do it. So basically, the higher the risk, the more relevant the bug needs to have. So, if it's a very bad security bug, we are going to accept more risks than just making a life of a few people a bit easier. I think that's a fair portion. Yeah, and I'll just say that if there are different opinions on what's okay or not, we have to let them release mailing lists, where I, of course, often say it's not a discussion list, but it doesn't mean that it needs to be a total quiet. Use the list to sometimes get an common opinion. Of course, nobody, even if they are part of the list team, is always right. We do make mistakes also as a list manager, but we try to learn from them to do it better. So, that was now for short introduction. I think at least one or two people now have something in mind to speak like France. I have two things. The first is a little note you said. If we do kernel API updates, we have to update the installer. That is no longer strictly true because the separation between the installer and the installed system is much better than it was for charge. Because we only use kernel maker packages now for installation. Basically, it does not matter what we are running in the installer for the installed system. Now means that... Now means that... When you install a system right now, it still installs the old kernel, and they need to update and we get a new one. That's what I've been seeing. It will take whatever kernel is referenced by the current maker packages. So, if you're currently installing testing, yes, you will get 2618 because that is testing kernel. Actually, I think that's one case where you will still get to all the kernel if you install from a full city build without net access and then update with some net. Because that's a new kernel to all kernel. No, I've been installing 10 systems over the last few weeks with net install, and I get a situation where I need to upgrade the kernel after booting. Net install includes the kernels on the disk as well. Yes. So, you will get those kernels installed before the security app source is added. That could be a big problem. But if it's a disk, that's not an issue anymore. Yes, that's correct. So, with point release, you will get the new kernels on the disk anyway. So, the next question is do you want to update the installer as well? And I think it's a good rule to do it if you have any changing update, ABI changing update. But there isn't the strict necessity anymore that we had for search. Because the ABIs were hard coded in some of the I components. Only on some architectures. On some architectures, yes. The use of the cases where it was, the kernel where it was. So, that's just an update on what we have to do and what we can do without. The second thing is and that's my most important request for stable release management is I would like to have all the information that's available accessible more easily from the release tab in the work page. Yes. Please update that, make sections for all stable and all stable. And add links to I can just tell you something. We are planning to do that. There's one thing that we would like to see and why I wanted to speak with this while that comes about we would like to move to the release tab in the work web page to the same machine where FTP master is running because it won't make our life very easier. But when I last spoke with some DSA about it, he said well, we need to check that with FTP master first. So... Well, last time it was it was still running on a spore but it was obviously a non-option to do it. At least... Oh, sorry, when FTP master was running on a spore. So, but now it moved so we could do that and then we could change it. I mean, at least just having links to FTP master website or something from release.dev.org would be good. The concern is that having it on the same machine makes it easier to do connections. Yes, it can be pushes extract data from the database and everything else. And at the moment you have very good overview of what is still to be processed but everything that happens after that is completely invisible. You have no idea what's having on build these, you have no idea what... why things are stuck there. I thought your room had added some functionality to has built the old few pages that you could see other suites as well, but I cannot find it anywhere. Actually, the other suites is not the list exported for some reason, because the security suites I have added. We need to export all the lists for the other suites. Basically, I think stable release management with the ability of information has been a last step, for whatever one or two years it's still a very large way to go. But for people who have to coordinate or check things, like I do it's extremely frustrating if you cannot get the information and have to ping people over and over again. Your room's page does have now basically just a pen Dis-Equals Testing or Dis-Equals Stable It's not documented anywhere I think. I thought I saw it as a form option on the page. I thought. I think we should just link such things to the list. Do you write that down? You realize these are pages that Ron don't need that org that are written in PHP so Ryan's not really pleased about that in general. Okay, so perhaps I should speak this line how we could add that information to the list.io web page. So Zepal points out on IOC that you could just use the list on local. Is that what you want to build status though? Yeah the main thing that we'll tell you now is approvals after things have made it through. You cannot get any want to build information from the project web database. So you can't get want to build information from the building web page? Not from the project, the database. The building web pages you can get, I don't think there's any links does he have any? No, there's no publish. Unstable is the only stuff that's published. You can even find links to build logs I think. Can you guess the build log? If you build a package you can find them because they are just listed in the full list for the package. Are they even stable? Not the textuality ones, one of the stable ones. But want to build the whole database status or anything you can't get? It wouldn't be too hard to get an export from Monabit or whatever at the two hours to at least the O page or so. Is that not so much as signatures? I mean all the rest of it is on building.do and that's a separate host that FTP maps for anyway so Just put it in the video. It's a separate host since built the moved from you left. Because built the state on life and FTP must move to Spore. No, that's separated. But I just said I will speak with Muro about it anyways because I'm not going to link a PHP page she's not happy about. I still have some PHP pages in my own that I use on there though. As long as it doesn't know what's in it. I need to work with him because he was working on replacing them all with CGI scripts and I need to work with him on exactly what the issues are that leave me going back to my PHP pages instead. And we will see. Okay. I just want to download some issue we should do. Yes please. Okay. Anyone else wants to say something? It's not necessary to be the mismanaged person to speak here these people seem to say what you would ask in the same world. So what are the plans for DI and kernel updates for the stable world stable releases? For the current one? Are we going to have new hardware support or are we just going to have security updates? For R1 we don't have anything planned the new hardware support will happen eventually. So like you were saying earlier stuff like adding PCIIDs and things like that will do an edge. And then we're talking about in January or so almost a year after Edge was released we're looking at adding a new kernel as a secondary option for Edge. The default kernel will still be the same thing. The main question is which hasn't been resolved yet to deal with that in the installer. Are you going to support new kernel or are you going to support both of them? Or are you basically going to say we only support for the installed system which is rather useless? Or you would say you can install on 2.6.8 and then upgrade easily. But if you need data access you don't have it yet. What you could do is take the unstable installer running the unstable kernel and install testing using a business card image or a network image. That would work probably. I would really like to see a stable DI installer which has a new kernel that's included so that you can just go to a new machine and install it. The big question then is that kernel is going to have very little testing probably. So what are you going to do with the questions? Because you don't want to have two simultaneous table versions of the installer. For one thing a CD image could never handle that. We just need one. I'm not too sure if we say we only add something like Netins to it. So that only one CD more or whatever. It also isn't that easy to install it because you would have basically two current first show. All your webpages would have to have dual links as well. It's going to be a mess. There are still a lot of other soft issues in that. Or the CD images? The surface fold. It takes a huge amount of this space to host CD images. It's increasingly clever about combining them. No dark CDs and so on. Is it normal anyway? It's mirror. It's a separate mirror. But a lot of mirrors opt in on mirroring the CD images as well. You basically have three types of mirrors. You have the iDrive mirrors, and the CD image mirrors. Do you have the raw figures of this space? No, I think not. Tell about it. I was going to... But it's really huge. We need it to... It's 20 or two CD's and 20 DVD images. Well, actually it's close to the CD image just plus the x-rays of the echo. Yeah, it's slightly over the fee. The idea is it is it anxious. We need it last time to ask you all of the stuff from all over the world. You need about one day to ask you stuff between servers. Which is really a huge number, yes? Only can get me to lie. Yes. It's really much stuff. Not a terabyte, but probably a tenth of a terabyte There might be a way to only have to change out the first CD? Yeah, it's almost not ready. That's not possible because CDs are built to the max. So as soon as CD1 grows a little bit, then stuff would have to be pushed to CD2. So they are built to the max at the moment. And you have to change the CD, there is a CD set up to be able to support that. But perhaps come back to another point, what you said about the... You expect several questions. What I would like to do, if you say we go with that kernel, that we publish the kernel in time, and say please test that kernel, a lot of users need to run it stable. Who might be interested in running a new kernel so that we have a serious amount of testing on that kernel before we publish it. Can you go through a back force or something to... We have an auto-build system, it doesn't do all architectures, but Wally has built server.net, there's auto-build kernels there, and the kernel we're planning on would be auto-build for CD, but we have to make some changes so we may set up the site. One minute, can we upload this kernel to the back force and then test it there or something? Yeah, we can do that. You can push it in a proposal? That's early. For post updates it's public, and these days you can really hear an experienced Debian user. You can really get run for post updates in your sources list. Even if you're not that experienced really. Well, you should know what the back side of the system is before you do it. Good opportunity to learn PDAP. Yeah, but it's now more or less modulated, so all this stuff goes in that's really accepted. That's a great improvement in my opinion. So I would push the kernel there, and perhaps make an announcement on GDA, that the kernel is there, please test it, and block about it, whatever, and then it will happen. So I think we could make sure that we get more testing. Of course other points are still to be resolved. Long as the installer doesn't support multiple sources for UDAPs, it's a real pain to test in places with a new kernel before everything's integrated into the table. We cannot say, okay, take from stable and from proposed updates when loading UDAPs. Can't we... Oh, okay. Do we need to make the proposed updates include all of the stable UDAPs? Then we would still need to modify the installer to actually look at proposed update, that would help, I guess. Yes, that would make testing a lot easier. That should be so hard, right? Really? And we can set the suite where we take UDAPs from easily. So if that's an option, yeah, that would work. Definitely. Well, we can put that in the next one. No, that's not so hard, because it's just a dark console suite anyway. So that's basically it. Yeah, for stable and old stable. I would have liked to have the stable and old stable updates out much sooner. Yes, I do. I don't think anybody is especially proud and so late. Yes, that's the least. Basically, Zovo and I started talking about doing stable updates and old stable updates really soon after the release of Edge, especially because old stable installations are completely broken at the moment. You cannot properly install old stable using the search installer problem. Yes. Yes. Especially since it's not that hard to fix. Okay. So the change then needs a DI update? Yeah, but we also want to get the kernel updates in for that because you don't want to do an R1 release and then an R2 release because the API changes again. Why didn't we get the DI update before Edge was released? Because we didn't know the problem access, to be honest. We suspected it would probably exist, but it's hard to test and so on when you're not in the situation. Yeah. So what timeframe is the stable release team looking at for between point updates? Actually, it would have been good to make the first point update something like six to eight weeks after the stable release because then we have seen also rough spots that needed an update soon. And after that, I'm okay with my personal opinion without discussing this further would be something like two to three months. But the license effects of most point updates But the first point update you'd say sooner than two or three months? Yes, because it's been two months now. Actually, I would say it's the first to be sooner because on the release itself we noticed a few rough spots shortly after the release which requires fixing also depends on how important is the point update. So if we say basically this time the point update just consists of this spinning or security updates on the disk I would say well, we do it when we have time. If we say we have important changes in the item make it should have to have a bit earlier. So on IRC says two months and the testing DI before point release would help. The IRC channels hash.com pop-up box hash.com-upbox.com So you're basically our IRC moderator. I was thinking more relay than whatever. I heard that one. I heard that one. I still haven't done it. I have no clue what you actually said still. That was my error. That's something like what I understood as well. I hope of the house in Harry Potter. That would be a good idea for next year. Just leave your rooms after. So you just sit up in testing the post after it's okay in your opinion? Yeah, there's only... I'm building it right now. I wanted to get FAA channels rebuilt against the latest source. But everything else I think is ready to go. Let's think about the TPU update. So we can just upload the TPU update. Oh yes, the TPU update. Okay, I just uploaded. What do you have in TPU? We need a new kernel in testing before we can let Edge kernel in. Yeah, in terms of what happened with that? What happened when you deployed the TPU new? Are you aware of that, bro? Well, it turns out that Neuros didn't do this all manually. But I already uploaded the Linux 1 version for Linux and that broke him being able to do it manually. I just talked about releasing it. Well, my question was, DAC used to have a check-in place that would prevent uploads of versions. I'm not sure I actually would. It did work. I don't know. What's the timeframe prior to the TPU news creation? I don't know if the breakage coincided or not. I tested that when I created the patches for the coping stuff to understand this. I guess in theory that also goes for the installer which doesn't have a new version in testing at the moment. It's extremely unlikely to get one. You can change that. Especially as Blasdian has now decided that he's not going to bother about migrating kernel for a long new rule. Yes. That's why my guess is kernel TPU now. Yeah, well. But we all know who we can look at for kernel migration. That's right. If it breaks it in unstable, I'll just reupload the old one to TPU and hit it myself. Right there. That's probably the best option. Yeah. I haven't looked at the modules yet. I need to do that and figure out where all the module packages are before we can really look at 2.6.21 on testing. Seriously? 2.6.21 is still broken for quite a few arches. Yeah, it could be a mainly... I'm not sure what it works on to take it. And there are also some build failures still for loop AES, for example, which would have to go as well. Yes, but build failures were related to the kernel itself, where they were... I'm not sure it could be build new as well, but there's a lot of stuff that's broken that's related to new kernel, new GCC and whatever. I thought we had the kernel building on our stuff for HVBA. Yeah, it builds, but it doesn't drop. If there's a bug that tells me that... I filed a few bugs. I think they were an RC. No, I didn't see them. I don't know. I think Steve will need to talk off about filling bugs with a light surveillance. There's an enemy behind you. But anyway, that's why we haven't got all the installer architectures on 21th moment, still a few on 18th. It's a kernel, been unstable. This is a shape I would like to be. It's a shape in a nice way. But as I said, a lot is due to tool changes. Yeah, tool changes are an important part of it. Okay, any further things? Well, did we cover the kernel now? I think I'll repeat some more stuff for example on RC. His idea is to do a point release every month. Old stable point release one month and then stable point release the next and alternate that way. But currently there's too much waiting for built ease and security packages not built on a lot. What shell are you, by the way? How did you write that? Hashtag DevCop-upper-upper-box-b-o-f-s. So the room rate is up a lot. And the app is just joined. And I suspect it's not a secret channel, so you can just go... I know, as you said, it's you, no channel is secret. I can't remember very well what you posted when Kuba. It wasn't me who published those eight mails. It wasn't me who published those mails. Whatever. I think it would be really the same thing to both alternating and then we would have a good situation. Not all updates are so hard to do, but we have some easier ones as well. I wonder if it's worth doing a full CD rebuild every two months for old stable? I think an update CD would be enough. So basically we need Steve here. It's a good reason for making up great discs. I know one company who has devian, which has a very wide policy because they are only allowed... or as a subcontractor who is operating the service is only allowed to bring stuff in, which is an official media from the... from the persons who are... from the publishers. Or as a subcontractor takes the full... That's not a concern for the projects, to be honest. Yeah, not so sure. So it sets a list, because we should have these updates. With updates you have a video you can put in. It's a video. So I think we should have these updates this enough for all stores. As far as CDs, didn't that mean CD already dropped the full mirror of old stable when Edge was released on grounds of lack of space on the mirror network? No, I think it's still on there. On CD? It's still on CD image, but not on the middle part as far as I remember. So we no longer have mirroring of old stable CDs? We only have very few places also. We just say, yeah. Because they said there's no way they could carry what sets all together. It's in CD archive now. But that's normal. One thing I would like to bring up related to release management in general is that I think it's too focused on getting the archive right and that some other things that need to be done for release are left too much to chance. And if you're release manager you should manage the whole and not the part. Such as release notes. Yeah, release notes. Actually, I thought it was a male boss. But release notes, we as release team definitely started way too late as release notes. And I'm really impressed again how good release notes and it is in the slaters. But it will be to make it better next time. The thing is that we have promised slaters that for R1, newly available translations will be included. And there are a number of bug reports about change that should be made in release notes. Which means that translators will need to update their translations. And you can only build release notes. Translators are reasonably up to date. And that process is not being managed. Okay, I will put that on my truth very high. Because I have seen how release notes work and I think I'm still one of the editors here. And the terrible creation has to be enabled again when translations are up to date and so on and so on and so on. I know it's a lot of suffering. Judith, I will follow it as one of my top items. If you have questions about it, feel free to ask me. But I'm not going to get involved. Yes, I can understand you. Yes, I will ask you. Well, if I'm a Glichtenheld, it's also one of the persons who know something or the other thing about it. Yeah, but keep it as a fixed item for future release as well. Yes, but for this one it's more important. It's more important. What I would like to do is just write down what needs to be done in which order and then put it somewhere so that people can do the same. How is the security propagation for stable working? The propagation back up into testing? No, the propagation from the security update getting into proposed updates for that one or whatever. My best, you just asked Zobel about that because he always complains. I'm missing girls. I'm missing girls on architecture like RMS. Yeah, I prefer Zobel's answer. He says that's working fine. Just be selective about what I listen to. No, no, no, no, you can stop talking. Because the elements of your often missing are going to be missing architectures. The beds in Beiser Bay are missing architectures. They really have an issue. So are they missing from security? Sometimes, from security. For example, what happens if there's a security update that doesn't contain all the architectures? If we accept it, send it in PU, new, then it will be out with both the security and the proposed updates and we end up with different depths in both sides, which is really a no option, I think. There's something we have to do about it. I'm not sure what the plight of all should be. Well, I don't accept that the build is complete. Well, to have it complete quickly and in security. Yeah, it's a complete, it's a great option. So why do you... Do you have any idea why that's not happening? I don't know in every case. In a few cases, it's, you know, people just trying to track down the build, it's more of this would be a better person to ask, but builds just don't seem to appear sometimes or it's machine down. It's also true that most of the security builds are done on machines that also do unstable. And while they are prioritized, and there's already builds running on more stable, like my thoughts, then you can have issues with the security build being released for some reason, on that architecture. Yeah. Or maybe by one day, but not one week. No, that's true. We've got to go for an office. It can be a week. I also have some kind of simulation, I couldn't prove it, but I have some kind of simulation that would be in some, maybe in some security team or five team administrators the communications are not working so well. Well, I'm not maintaining a build even that actually also does security, so I'm not really speaking for a time experience there. But usually when the, when a security build is done, security people actually get the build log as well. Yeah. So, and they, I believe it's even in them. And I think communication is the case like, oh, why didn't the build up here? So, what you usually should do, or what I think, but the reason that this thing is so true is to write a mental build up, or whatever it is. They usually do that. Yeah. I see it sometimes on my own. Oh, okay. That's not the issue. But yeah, sometimes, obviously, if you've got a build running on stables sometimes that breaks, or, well, that shouldn't break the stable builds, but if you disable your build even to fix your unstable build, that can also... It's almost a thing, yes. That's correct. Right, so there's this, it can interfere because they're almost same host. It shouldn't happen, but yeah, sometimes it does. Yes. I can see for the members, I think some such things like build teams will actually disable because something was totally broken on stables. And then after asking them about stables, they're enabled only for security. Yeah. Yes, but such things are really painful. Yeah, sure. But it's also usually an exception, I think. I don't know how often it happens that a build doesn't break the security. Almost always. Almost always. Okay, so then this should not be the week. Then something else is going on. Sorry. Well, Martin just said that you can... I think when you have lots of issues, like with arm and spark, sometimes it's M68K. Yeah, I don't know. That's usually because M68K is built in a normal stable and that takes a while. But basically we shouldn't... We have to at least see fast unbuildings. So we shouldn't have issues then. Well, there are issues with arm currently. I've been working on the arm mail list. And they have, I think, four machines doing building stuff with two being offered to run as a building for months. And nothing happening here. I mean, three months. Well, months. Yeah, months. Three months, six months. Okay. Bill Gateliffe's machine has been an offer that's been outstanding for over a year, actually. I don't know anything else about it. Actually, it's a machine which is... Bill Gateliffe, he's the guy that's a 600 megahertz machine whereas the other ones are 200, 300 megahertz machine. Yeah, we literally accept that machine. Sorry? Let's see if you should accept that machine. Yeah, I think so too. But I mean... I'm not a part of that. Please, on the arm mail list, I think that's a really problematic situation that people don't even know why it's not being accepted, whereas that's a post-modernist and I don't think that's a great reality situation. Wasn't hosting the issue with that specific machine? No. The machines that hosted some... The owner is offering hosting. The owner is actually not a DD. I don't know if that's a factor. There are... Well, I mean, there are... There are other machines. I think most building machines are hosted by people who are not DD. And if it would be an issue, because it's certainly not hosted by the DD, by the partners Most buildings actually do have a local admin who is a DD. Yes, but there's also... There's usually also a local admin who's a DD. True. But I don't actually know what, for sure, what... What I'm saying is just... I don't think that's the reason. The main purpose is that you don't get the information about why. Exactly. Otherwise, you wouldn't speculate about saying the reason is that... Right. So that doesn't matter. I know there's also a request open for a very nice offer of a spark machine. Joyfully. Yes. You'll support it? Yeah. We have enough sparks, basically. They do have... We do... We do... Where? Any active or you may offer? We have... We could just set up two more induction if you want. Well... There's only one active, basically. The only live spark right now is sparking. Yeah. I don't know what happened. The portor system for spark has been down for ever now. Yeah. And the... Or it has been down as a bill deed for two months now. When? I don't know where it went. Steve? We have enough machines there. We could just... We have hosting offers. We have IP addresses. We are now setting up next... We are just about setting up... You are talking... I take what is a machine there. You are talking about available hardware. You're not talking about... Configurability. Configurabilities. Configurabilities. Or machines that are actually being considered by DSA to be added as... I'm not speaking about the last slide. I have offered machines there more than once to DSA. And we were talking about machines actually running and available for work. So if you then say we have enough spark... We have enough... Good offers for spark. That's what I meant. Good means hosting included. So you should file the request tracker. But still... I can do it again. Make it visible that there aren't offers. Yeah, because they only offer all of the... It's joyous. Joyous. Yeah. Now as far as ARM, today I believe all of the buildings are back up and running. All of the buildings... That's Greek. Toffee. What's the other one in that class? We have three fast ones and two slow ones. We have three slow ones. Okay. Elara, Europa, and Netwider. Three slow ones. They're six slow ones. Am I not located? Greek. BDL Facebook. Oh, okay. I missed that one. Netwider is netwider.gag.com. Yeah, I have two measures in the face in this one. They were disabled for... They were disabled... They had been disabled because of the issue where they were running a kernel and it was too old to be compatible with GLC. That's been resolved and as far as I know, where all three of those have been reactivated now. That was basically a good toolchain issue again. It took a while to get it going, but as far as I know, they are up and running now. So, in terms of where ARM is, in terms of power, it has been catching up with the backwater. Is that power processing machine that was moved? Is that back up? I don't know. I suspect not. There were... Snowman is then online and talking about it. Yeah. It's not... There are some specific issues with the hardware and with his local hosting that... Yeah, there was a problem with the mail... Yeah, there was a firewall issue and the second problem was when Ryan was looking at it one night, he was poking around in the system because he was trying to figure out if he could take Voltaire's kernel and put it on mail out so that mail out could build up safely again. He looked at PROC CPU Info and PROC CPU Info was reporting temperatures on the processors and one of them was like three times the temperature of the other three processors. So I was like, oh, is that a hardware failure? So I think actually Voltaire was... Last time I knew it was being taken offline so that Snowman could look at that, but I don't know. I don't know if it's been number nine. What's that? It's responding to PMS or something like that. Is it responding to PMS? Is it the machine? Is it currently in the boxes? You know... Yeah, I don't know if it's been... doesn't seem to be accepting any of the edge connections that could be as firewall talking. That's true. It could be the CPU burning. So, but yeah, Melo is up and Melo wasn't building on stable for the longest time because of another kernel issue, which I didn't realize that had been the case, but yeah. After managing to walk himself out of it once, he got the local evidence to agree with that and so Melo's been building now and so RPC is keeping up with one and both here is somewhere on the edge. Okay. How do we get into this whole discussion about build design which we're supposed to be talking about because we always end up just a bit deep. But I don't think we're going to solve anything about build design here, are we? Okay. Unlikely. Well, we've got lots of luck out there. Yeah, we've got everything like QD, we've got it up. Talk to Schlatch about CD releases for stable release. Yeah. What to do with them. Whether to just do incremental update CDs and when to rebuild, we should have a policy for that. Yes. Sorry, technical play in this pop session is meant to be urban now. Sorry. Okay. Now we can switch camera off and we can just talk about that. It's 12 minutes of tape. Bar open, yeah? Okay. Do we have an other topic that he's like as a perspective about can we do a 307 this week while we're all here? Is that possible? Can we do the next search update this week? At update? Yeah. Search. Search. Yeah. Search is a help of I cannot upload the I yet for something. Okay. Because I have I've thought so much to look into it. Sorry. Please I cannot see anything on the page. Why? Because it doesn't list stuff that's already been accepted. Or we're talking about a completely different page that's possible. Are we talking about not being able to see one of those status for things that are not stable? Yeah. The only thing I can do at this moment is to do our medicine basically and check what's available but I cannot see anything beyond that. Yeah. And it's it's been unchanged for at least a week and a half the same architect's missing. Missing. So it looks like nobody's paying any attention. So do you know one of the packages you have provided? Choose mirror. What's happening? Yes. Root scale is another. Zoho pinks Elmo body yesterday so something may have happened. Choose mirror is being built on so Bayesian stories. Choose mirror one of those status I see uploaded or needs built for all arcs. Okay. Old stable. So that means if it's been accepted it's not clear. Back to one bill. So there are always needs built M618 needs built. Choose mirror HVBA needs built ITRAIL 6 Is that actually a bug? The question is why it's taking so long because they have been there for the past two weeks or something like that. Yeah. Well they're all sitting in needs built so they're not building so they haven't been picked up at all by ability. Zoho says it seems that there are a few built days not configured for old stable. That's what I was going to say. Regular. It's our being built aren't it? Yes. The security is different than this thing. That's two different Yeah. What is all state security? We have our state security now. So I think I think that's my issue. I think the state will want to build it with just the needs. Okay. So this needs to be fixed of course. Yeah. And that's also why I've been trying to push it to Zoho instead of trying to fix it. I don't see why I should have to run after that. Choose mirror as a package. Choose mirror, base installer and root shell. And I think those are the any packages. You know they are many packages I upload. That's missing on M68K on ES64 on SC90 and Spark. The same problem free. Yeah. Just put it to make sense. Are we going to be kicked out by the next block? Yes. Peace guys on the next block.