 I usually, you know what I'm saying, exactly that. Right, but Paul has worked a lot of way against, he has demiannied a completely separate branch, said something about stream source in that branch, right. Convince me, this makes sense. You're crazy. Let's make it easier, huh? He also said he likes Issy and Bill. Yeah. Issy and Bill are great. But, you can, like, you just, like, get long on the top there, and don't worry about it, and, like, you don't have to worry about it. Nice job. Nice. And when you have, like, you're pretty good at those reasons. Reasons. Because of reasons. Reasons. Wait, reasons. It's not at all reasons. And, um, I have, like, a whole spiel, but I, like, I will personally write docinfig.com. docinfig.com is for you to log in that directory. Yeah. And you never end up, like, even, like, when you have upstream artifacts, they can, like, get out of sync with the tar balls, like, you forget to pull, and, like, the packaging changes you build in, like, as a core game, and then, like, it didn't happen to us yesterday. It didn't happen at all, did it? Um, and... Wait, actually, I don't understand. Well, for me, I like to... If you, like, forget to pull correctly, then you can end up with... If you get to do the merge from upstream back into your Debian branch, then everything agrees. Well, you can do it later, but my problem is... But then it doesn't, like, it's not correct. Right. For me, for me, I merge from upstream if my, the tar balls are uncant contaminated. If, if I have to do Debian free software guidance and repackage the tar ball, I forget about the upstream history and I generate my tar balls from their upstream distra, uh, commit history, but then in my packaging, give repository, I commit just my clean tar balls and then I go off that. So, for me, I do both workbooks. So, once you have generated the tar ball, you commit it to pristine ground dimension, right? Yes. Yeah, the ones that I generated myself, I commit those and then my history only has, like, Debian packages, patches, and, uh, Debian free software guideline tar balls. One per commit with no upstream history because it has all the bad stuff. I wanted to ask you if better tooling could solve your problem. And, for instance, if we had a pre-received hook that made sure that the upstream branch would push before Debian branch and a Debian changed log entry of 1.2, actually descended from a tag 1.2, that sort of thing, wouldn't that? That would make me a lot happier. Yeah, yeah. That would make it a lot happier. Because you can get those two hops out of sync. I still think it's, I still think it's, like, it's still when I see it, I, like, I don't like it. But it's fine. I can live with it. I have to work in other teams that do it this way. It's not a problem. You can filter which tags and grunts you see. Yeah, yeah, yeah, yeah. You, like, look at the get tree in history, and it's like, ah. So where do you need to, um, where do you need to, uh, strip things out of the world? If I can't redistribute. Why don't you make the, no, no, I said why? Why don't you do that? Yes. Why didn't you make that, those changes, as I commit on a branch from the, the upstream? I can't redistribute that file. That file is part of the get history. I am distributing get history on ADIF. I do not want to distribute that contaminated at history. That is a great answer. Yeah, but that's a non-sold answer for all packages. Anything on ADIF? If it's not distributable in the architecture. I mean, if FTP master rejects that file, why should I store it on ADIF? Yeah, but that's, none of you can understand how about what you've got relates to all types of things. So what do you, what do you... I would agree with these draft points, such that physically it's separate, but there is a link in history in me today. You often said about things that are not distributed by the person known for it. Yeah, right. Not distributed. Yeah, exactly. So that seems like you can solve that with a, just like you have the... Is a tool for that to get, like, yeah, draft points, yeah. Then you get it in history. Yeah. So the same, the same sort of thing you do when repacking a source in Demian rules, you do that to every single commit. We are in string, and you make a... Where what? Where string, the room speaks to the image. Oh, God. Do it. I'll call you back. Hey, York. Hi. And like, none of the media teams is here, so we couldn't tell. Yeah, move to a camera. There's no reason to turn it down. There's no reason to turn it down. Yeah, that's fine. It's fine. Yeah. So you couldn't do something. Yeah, I'm proposing tooling, and I think this is not a terribly complicated tool and correct, where you say, it's my point to make my Demian history diverge from, like, commits on top of my non-Demian history that I can't stir in an alley up, and the parents are there, but there are rewritten bits that really exist on the alley up. So you can clone it, you can actually interact with it, but some of the parents don't exist. Well, I think there are two points. One is I'm quite sure it's a non-enforced policy right now, and I'm sure there are loads of undistributable stuff on either. So there is no particular reason why the patent, the ensued care, and others not. Okay. That's one point. The other point, I think revases are okay, and I'm doing revases and non-enforcible things on various of my repositories because mostly I know I'm alone. And I think if you may have the maintenance list saying I've revased for this or this reason, please make sure you have the new version, then I don't think it's that bad. I mean, here at DPM, it, in essence, does revase each time, but it keeps the history of each revase due to mergers, such that I can go back in time and see how the patch series was revased on top of each upstream. I mean, yeah, it's simple. It doesn't say anything. And in that sense, I kind of like it because you do want to see the clean revase of what DPM is going to package here. You do want to see the snapshots of that action. DPM is an excellent example of this, but it's worth thinking about this conceptually, that even if you are revasing commits, if you have a different branch that has each of the results of the revase as one of the parents, now you have a non-revased history that puts this. Yeah. And that's what you think. Yeah. You should get focused on things we have to talk about. Yeah. Okay. So, we're seeing now, on the workflow, you see that everybody is okay with using Prism Tower. Yes, yes. I've been using tech-based generation of glory, right, when I just follow the should gates. Okay. Is that, are you comfortable with that, or do you think I should? But there are a bunch of Prism Towers. Is the git tag signed? Yes. Do you verify the signature of that tag? I do, and I have the signature, I can sign the signatures with upstream. That's fine. There was discussion in the previous room where we said we should be using Prism Tower, even if we are generating our tarp balls, just so that we know how the tarp balls are generated and you don't have to guess. Well, I think the answer to that... You don't need to guess because there's the dgl and gbt.com that tells you that you are using that action. That's the bullet. I think that's only the case where there aren't upstream buttons. Okay, that's for you. I prefer to use the upstream gates. Yeah. Okay, there's that case, so you will have to address it this way. I mean, what to do if there are both... What does I prefer to use the upstream gate rather than the tarp ball? Are you guys okay with it? Are they both identical? No. Usually not. No. One point is that I'm going to use exact permission. But what is it... The problem is that the gbtag is signed and you can compress it better. The tarp ball is often not signed. Well, this is my question. Is this the open stack sign? This is open stack sign. She didn't make the problem solved. No, no, no. And they... They didn't make the problem solved. No, no, no, no. And they also signed a target view that was like that. Yeah. Yes. But that's not my question. The question is, is it okay with you guys that you do like that or do you prefer that they... If the original tarp balls are published, we should be using them if they are signed. No, I think that's current practice. That's current... I do have concerns about not using a pin tarp ball if upstream is generating tarp trees. I do because it's less efficient. You're going to spend more work. I'm less concerned about the efficiency than I am about the discoverability of what the packaging is doing. And if you're packing... If we're working on the same package, okay, it's a team package. Yes. If you haven't done it this tarp-all way, then I come in and I say, well, I haven't seen Thomas in a couple of weeks and I need to fix this thing. And I do what I think the team does. Then I'm not going to know that... I mean, maybe if you specify everything in the review.source. No, no, no. That's fine. The question is that I will have to do that to do a review.source, because upstream do want me to document that. We're generating tarp-alls. We're choosing what the production is. Second thing is that, as I said, in the dvm-slash-gvp.com, you can say I'm using tags. So there's two ways you can do it. Yeah, but it's still something that is less discoverable. It's a deviation from the norm. And I'm not saying that it's not a good reason. I'm just saying I'm uncomfortable with it. And for me the problem is, for me the problem is a lot of times the tarp-all is different from the history, such that when you do the export, it can be different from the tarp-all. Which is why I want to use it, because usually it's clear. It's not clear. The generative files are there for a reason. I mean, I've seen things that... Yes, it's not for us. That's what they use as tarp-alls. That's what they intend as things as a consumption. As an upstream, if you're releasing to the Chishog, your mental image of that release is that tarp-all. Yeah, you should make sure that those things actually built from source. Yeah, whatever it is in the tarp-all, a lot of times the branch builds and the tarp-all on Chishog does not. But that's true. And I've had a couple of cases like that and I go back to the upstream and I say, you act up your tarp-all. I'll send you an email. And that's the good thing, because it makes life better for other consumers and totals. So it means that no matter what, we just use Chishog. I think we're never going to get to no matter what figures or might be an option. I mean, if there are no tarp-alls published, we don't... Yeah, yeah, yeah. If there is a tarp-all, we use tarp-all. If we publish them, we use the tags or commits or... I have another question. Of course there's little cases we have to rebuild tarp-alls because upon briefing it's all that you make an image in there. Or is the audiophile of how we pronounce the name of the project? We have techniques in deviant source to do that tarp-all. I'm even a bit of a pristine term. So it's the same 90-10 thing. 90% of the package is you should be using upstream tarp-alls if they exist. Otherwise you should be generating tarp-alls and checking those into pristine tarp-alls. Yes, that's correct. Can we put it in pristine tarp-alls? Yes, yes. That doesn't mean you cannot use upstream to get history. As upstream. Either pristine tarp-alls have a large with whatever was put in the tarp-all then. Or you create another branch that is like upstream plus artifacts or upstream release artifacts and you just create the dips and then you have something and I think it's more interesting to have upstream to get the repository than the only tarp-alls history. Can you get the individual changes? Branch names. Do you want to start with that? Yes, I'll do the thing. Deviant slash thing. We also have to talk about tag names. But for branch names, I mean, I don't know. I didn't have a lot of experience with it but when I used not the default branch names, I thought it was a good idea at first and then I kind of hated it. So then I renamed the branches to it's master pristine tarp-strup. Yeah, that sounds good. Yeah, that's not good enough. That's not good enough. Yeah, see? Okay. There's one between you and me. Oh, sorry. Text is in just so relevant. Not. We're seeing tarp-strup in master. So why doesn't it not be in master? I prefer gaten slash said open to slash trusting. No, master is master. Master is the head of development. Master is the branch for a stream, never use master. For a stream. I like that opinion. You will have a head of trouble if you rename it because it's a stream. Well, you will add... It's an origin slash master. Yeah, right. What I do is master for whatever is the head of development whether that goes to experiment or to sit or do not do whatever and then you do a suit-based name. Master dash wheezy if you have to do a staple. So your rebounds are origin slash master which is devian slash master. And then your remote is version slash master. Your upstream, the project that is publishing the Python without any decoding is a remote named origin. Origin slash master is their master. Us, we would be remote named... Why wouldn't it be origin? That's the name. What you have on alias is a representative with master that has the devian upstream that is whatever is master on the upstream side. And how do you do back ports and what would be its name? Wheezy back ports. Do we need master dash? Whatever. Master dash. You can't ask anyone what master means on the devian packaging branch. You will have a different answer depending on who you ask which is therefore why I think master's name is wrong. I turned the question differently. If you ask anyone what master means for a key to a repository they will always say it's the head of development. Who's the developer of your... They were repository. So... the upstream is obviously has much the same history but these are two different master means. That's the question. Did you say master which means two different things? No, that's fine. It turns out that master means different things when you check up on an entirely different repository. This is not surprising. If I have a key to a repository and I'm doing the work on it it will probably be on master but it will be github.com.jp slash repository slash master not github.com.jp That's fine. That's how people tend to use github. When I created the poor world slash linux on github and it's xnox slash linux and then I have master branch I don't think anybody believes that I actually released linux models. Yes. It's better if you do a dog use of that. This is one of the things that github when you click fork, you keep the description and so I'll put your ringos and say the official mirror of an amygdala. That's true. There's still another way that I believe that we should make sense of our branches and that we should have branches targeting the distribution of that which I think is the best to have the beyond same thing because is master pointing to experimental or to a seed. Where does the 99% of your development go today? Trunk. Get out of here. That's true. Virtually all of the development is really good and unstable. We'll change it a little bit. It's good. You can. You will end up having to write that in a bunch of places and it will be tedious to write. That's it. You have to do it on bts-starfield to know which thing to check out. We'll need to update the remote head of every single repository. It's already doable. I think the head should always point to a seed. Yeah. And the easiest way to make sure that happens is to use the name master. You can also use the name master. We probably will one way or another. There is no way to get protocol to change where a remote head points. Of course there is. I know. Maybe not. Using SSH, yes. But I'm saying that for annoying reasons that really didn't get protocol, there's a convenience factor in remote head being made faster. And I think if the problem is that upstream's master has to be upstream in an hour, repository is an issue that's just either a tools or documentation issue. We should not develop like that. I also... Well, I guess, okay, we'll call it master job stuff. Well, then the question will be out there. And you start your own packaging team. Oh wait, you have. Oh. Oh. Burn. Oh wait, we're streamed. It's not necessarily SID. No, it's SID. It's practice. We actually know which one we're talking about. Let me find an example. You have released a new thing to SID. You're now doing some weird changes on the packaging. You push that to master for comments and then someone says, oh no, you push that to master because it's the head of the government. And then people say, well that should probably go to experimental first. So you cannot roll back the master branch without sending a message. So do you see that you are sending me that you're going to check out a repository that will never know which one you mentioned? What is it? I like how Digit deals with that. So it has a set of branches which must match what is currently in the archive. Such that if you get an NMU and you know that you take that branch you commit an NMU and you push it. And then the head of master development is a separate branch. Most of the time it's the same commit as the Digit and SID. But at times where you push something new which hasn't been uploaded yet you push it to master but you don't update Digit and SID. And then once you upload and tag it then you match your master and your Digit and SID matches. But that adds an extra branch to maintain. But that doesn't work all the time. You have Jesse and you have a package in Jesse that matches the Digit and SID slash whatever. So the suits don't match. What matches is the version all the time? As far as untyed of course. As far as Devin and SID, Devin and SID do just watch out for influence and mismatches. Like I think Git and Pianist are the ones that you end up using there. It still has some problems with master style branch types of slashes in them. I think it produces it works but it produces something very untyed. There is an assumption that Slash separates the remote from the branch's name which is not true. I think you end up with upstream dash Devin slash SID or something. Is that one of the they don't use slashes in the tag names whereas tag names are strange. I forget the exact constraints. I wasn't sure what you just said whether you meant remote Devin and branch SID or branch Devin slash SID. I imagined it from Devin slash SID. That is the obvious confusion. So I would suggest dash. Either master dash or just go well over time. We already have an hour next week. Oh, did you? Literally by SID. So we agree. Make me a free branch. What's that? Yes! Yes! No, that's great! So we have two choices here. Either master dash or just we SID. I suggest master dash we SID. For when you do a stable update. When you, okay. I prefer not having master as an element. Just throw it on the proposal. Either A or B. I was proposing B. Yeah, just do wheezy. I think it's better. Just wheezy, yeah. No master. And we have a stable guarantee that the names should be defined. Yeah. On both sides. Sorry. Yes, we do. So you wouldn't because they're in that. What else? What are you just using? Get DPM and give them packages. Get DPM can have you ignore it. Sorry, you can ignore it. So I think that's a problem. So what is the type name that DPM is our DPM slash the version number. Version number. Yes. We want to do the default for this one. Yes. You shouldn't really not have multiple branches or upstream. How do you do for following wheezy and you're well. One of them is a stable release branch. So for multiple upstream branches, you obviously have to evolve the history present. You don't necessarily have to have a push wrap for old and old time. So the project text is to stick the of the commit that is the upstream for that branch in DPM slash.it along with some other stuff. That means you can always reconstruct the upstream when you need it. So no other... Just one branch. Because in the previous hour we decided that we have the full history. So if you have the full history your DPM branch has history in whatever version you're based on. You have the third history but there's no reason you have to have a for the upstream and a branch in alias. The fact that it's part of your history as long as you... The objects are on alias but not necessarily behind a branch. They don't have a branch name associated. I have a question. Why do we need an upstream branch at all? Just keep upstream tag in the DPM history. It depends I don't know what the DPPQ is. Right. DPM was constructed literally when it didn't get prepared. So there's no need to... I tend to push it anyway. Just from zero. I have no idea what the other sort of is. And the fact that you have a branch for things like upstream is quite common. So it's an expectation to be able to like you block the slash upstream. I think that makes sense because there's an obvious single branch for upstream and in the case we're talking about experimental is using a different upstream branch or stable is using a different upstream branch or whatever it is then it's not clear what should be the upstream. There are some... You can always... As depth commit generates as depth commit generates Depth commit does that? But type it out that we base what depth commit does instead of actually specifying. Depth commit dash R. It finalizes your change log. It commits it and it generates a signed tag which is suitable for a git if you have columns or weird things in it. And then you just push. Does it get signed attack? By default it asks you to sign. And what is it? Depth commit works with git, bazaar, mercurial, SEM, it works with anything. You wouldn't do that on upload. You wouldn't do that on... Exactly. That must be before upload. Well it is before yes. Because... Minus R gives you the upload commit. Well... For depth intact before upload. Before upload... The match the upload. Doesn't really matter. Corresponding to upload. Corresponding to upload. Does anybody know if it's possible to get the pushed used dash dash follow tags automatically? Yes. There is a conflict option in 2.0. So the answer is currently no. What is dash dash follow tags? Follow tags is push all of the tags that are referenced from something else, that refer to something else that is pushed. That is in the target. It's what you actually want. That shouldn't be anything else. That's like... If anything is new, then push the network. If there are any tags that refer to objects which exist in the thing you're pushing to, in the remote, then push those tags as well. Because if you do git push tags dash dash all, it will push all tags that you have which may be completely random crap which you've tagged for yourself and shouldn't be public. Dash dash follow if you push master and the tags that are reachable between now and then in master all of those tags will be followed. It's not because it's in some work that you might want to do things. This is where I started this thing. I think for the work loader it's too much human. We didn't set on. As I understand we can use inbuilt package in git epm or git qp. That's git. On the first and the second option if one or the other doesn't matter then it's just inbuilt package. The question is do we want the third one? Isn't that question whether we have to specify a helper tool or we can leave it up to the individual maintainers. The latest things I've seen on the mail list are that we can't use git tpm and git build package pq on the same package. One leaves dpm leaves patches of slides without documents. They don't use built but it uses these patches of slides and also I think it requires pushing a separate branch that has patches in it. They're not only incompatible they also use totally different parameters which makes it even harder when you just lend a new package. I suggest we should set them on one question. Can we automatically convert the answer is yes because we need to actually use the answer is yes because we automatically convert to 3.0 slash build from the tree which is the dependent and we don't want to expect to come out and say just that. But they both generate the source privilege of the representations in it. The tree itself. Can you filter branch one into the other? Is it possible for people who want to use don't say filter branch because it's destroying history. If the team wants to use git tpm as the canonical representation and I want to use git tpm is there a non-painful way for me to make that happen? Who has used pq? I think almost no one has used either. I try them both and I've also used git tpm just that on that one package the six package maintains. I felt more comfortable with git tpm but I can't say that I have a lot of data points on that. In that case, because everybody never tried except you then you are all in the case one and two. There is still the simple just have a git history with the git tpm slash batches manual management and then just do your quote statement. Which exists but it's a work in a little world. I mean, if people don't have enough data yet then I don't see why you need to decide this not for truth. I think that's a great strategy but also along with that is that we should with two things. We should not do a mass migration until we do decide on one or the other and I'd like for there to be once that decision is made even if it's a painful way to convert A to B. Is there any way to import the git tpm it's certainly important to git tpm it doesn't import the git history but it will take current state turn it into series of commits against upstream merged into master so we could import the current release it from snapshot with quote series later switch to a helpful one. How are we converting the history? Those were dead. Are you keeping SVM history or are you just importing DSC as it is? I mean what would you decide? The tentative plan was to import the DSC of old stable and stable and don't go back to snapshot if anyone cares they can use git replace so just one for old stable DSC for old stable stable and so just three three commands three commands three commands we can generate the whole thing for git tpm the whole thing for bq and tell people here is a sample two repositories go play around with it and within two weeks tell us which one you prefer and if it's just three commits then we can convert it using either of the tool and everybody will have package that they know to play around with it For that sort of thing you really only need to convert it instead for something that has a few patches yeah oh yeah packages without patches they don't care for them well yes the few is better than having a hundred to ten patches yeah it's just you just be thankful I have a source for my one package sorry I have a source for my one package die a fiery death die a what? a fiery death but that's sort of the same but so are we saying we're saying doing a mass migration into Bode or just opportunistically I think it's like 10 well realistically if we want to have some results we go to set up a deadline which one we decide which one we want to do so how much time do we give fifth of November is a good date after that sure yeah well that's actually a good point maybe we should do that for experimental after everything that was because the important work now is making sure what we want is increasing so we might have some preparation work and do that after the freeze and prepare that colony after the freeze is done I think that's a good one because otherwise we will have half of the freeze within Jesse yeah the fun part is all what we should be fixing most of us you're making a point that we should migrate before the freeze that doesn't mean we don't prepare the migration I think we should prepare for our permits but not change anything on the SVN in time for Jesse because the energy should go into Jesse and not into the end so packages are what it just did before you should build the SVN this means that if you modify package after the freeze that's the problem of how source packages are currently I mean how packages metadata is per suite and is uploaded to the archive that means when you want to check out an old stable package someone should have made sure the representative is still there so that's a job for I have a slight refinement maybe to the pros which is um that I guess I could say and I don't know if this is a good idea you could opportunistically if you were interested in trying one out you would opportunistically convert it to get and just maintain it and get using one of the two helpers one of the two regimes and essentially ignore subversion from then on if the regime that you chose is not the one that we end up with then we'd have to convert that particular package but we're not going to do a mass migration until after the freeze is over and that allows I think that allows people to experiment but not have to worry oh my gosh I now have to upload it from the GIT repository but now I have to also keep subversion from the GIT if you do upload from GIT you can change the VCS yeah absolutely I believe that we will have a lot of activity in our VCS after the release and I think probably in the freeze time we're not going to do the migration yeah yeah I'm just saying that from now until then I don't want to have to keep GIT and subversion so do we agree that so we keep the energy in the freeze and then the migration happens in the freeze is that sorry say that we don't change anything until the release of November and then after the freeze we can stop the migration the mass migration until before the mass migration you can pick a package because I think that if you're going to say I really want to experiment with GIT TPM I'm going to pick a package that I'm working on I'm just going to do the migration I'm going to learn how to do the migration again I'm going to figure it out I'm going to learn all the pain points but once I've done that I'm just going to live in the GIT world I'm not going to deal with it I'm going to update the VCS errors and I'm not going to deal with it can you continue to use GIT yes because then you just use this again and the same so you can do it so your proposal is that from now until the freeze you can use either SPN or GIT as long as you're using this GIT workflow you can use some GIT workflow that goes by those constraints that we have decided on but it's free to pick other stuff that we haven't yet decided on then you also have access to many packages don't go around converting every other just because you like it but if you would like to experiment you're welcome to when we consider team-maintaining we upload it to CIT we get it into Jesse and be ready to do that possibly we'll because we would rebase SPN anyway we're doing Yu-Gi-Oh! technology I'm even cool with saying pick one of each in other words if you want to go switch to GIT pick one and do GTP and do one DPM and don't convert anything else just pick one and do both and see what you think and absolutely let us know email me when do we decide then number one number one some sounds involved number six leave a few no discussion shouldn't already happen before we need that deadline make a decision before we end number six maybe a few of us could do some conversion stuff yeah would you say on December 1st December 1st number six that's what end of August so October yeah is that enough time to experiment it needs one or two very motivated monkeys to do that anyway unless someone volunteers to be that monkey it will not happen to them November 6th is a decision deadline nobody's planning on getting up warning December 6th migrating everything by yeah anybody who wants to do the migration you can come over to my house and where are you I'm in DC I was left to Maryland I would go with that yeah so wait a second wait a second I must come yet that I haven't checked with my wife now you're trying to get happy everything is it's on the right it's on the right it's on the right it's probably gonna end up in there I live out of a hotel actually it's on the right it's on the right it's on the right lol cool E-mail lists that's somebody yeah big question add some of the E-mail's I can E-mail's I can it's your name that's good thanks buddy I have a seat I actually said that when I'm here. I think you have to say yes before we leave. I'm prepared for the shot to do it, but I was going to and I was supposed to do it anyway. Okay, cool. That's not so commitment to do it. All right. Well, you know, there's a lot of hacking time here, so I'd love to sit down. We can try to do a few conversions. I don't think it's too good. All right. Woo! I wonder what about kids with a scum sack? Wasn't that kind of the point? Yes, sir. There's so many fascinating things back here. That's my one. Well, fascinating and stomach-turning. The fascinating ones are my one. Yeah. I'm ready to go. I'm just kidding. I'm ready to double. I'm ready to double. Honestly. You know what I'd be excited at? Writing cool. Automatically takes you up a lot to the cheese shop. Turns them into pristine tar branches. So there's just cheese-shop.vn.net, which has a different repository of everything that you sell. Like packaged. Actually, I could just auto-STDM it. Oh, no, no, no, no. There should be apt.gshop.vn.net. No warranty provided. Do it. For all your untrusted codes. Hey, Docker. Right. Don't run it as root, though. Well, root is Docker. You can't do it. Yeah, it doesn't look like it. It's not true. It's not true. It's not true. Well, it doesn't make me any more happy. It doesn't make me any more happy. You would actually do that. We're gonna see our next review in a few days and you guys can track your safety data. What are you talking about? I'm not sure if you're gonna do that. You're gonna do it. I'm not sure. Why would you do that, if you want to do it. I'm not sure. I want to see it. I'm not sure. I'm not sure. I'm not sure. I'm not sure. Why would you do that. I'm not sure. I'm not sure.