 Welkom bij de deffroom van De Deff, de laatste afspraak van dit jaar. De meest vragen over Deff, is natuurlijk, wanneer we verrelden? We zijn in de verrelde manager van Deff. Leuklaas, en hij zal ons een beetje vertellen over Lennie. Leuk. Ja. Nou, vloon is niet echt de wereld, want ik droog hier by car. Het is alleen maar een uur. Dus ik zal een beetje over het volgende afspraak. Het is namelijk Lennie. Eerst ga ik een beetje over de nieuwe afspraakteam voor Lennie. Ik ben een van de verrelde managers. En Andy Bart is de andere. Ik weet niet of hij hier is. Over daar. Hidden, ja, hidden op de rijk van mij. En dan hebben we nu veel verrelde assistenties, want sommige van die zijn gedeelte. En sommige zijn hier zeker, zoals Mike. En we hebben natuurlijk nog steeds onze verrelde bezoek, Steve Langaszek, die is nu meestal bezig met boontoe, maar oké. Maar we kunnen nog wel op hem gelaten, als we vragen hebben of er een grote binnen-en-en-mu zoiets is. Dus, een van de dingen, we hebben twee dingen eigenlijk voor verrelde om te manageen. Nou, technisch. Namelijk verrelde critical bugs. Er zijn op het moment, 500 verrelde critical bugs in testen. Testen is de staginggebouwen voor de nieuwe verrelde. Ja, ik ga er gewoon in, als je een vraag hebt. Dus, 5 RC bugs in testen. We hebben meer in onstapel, maar natuurlijk zijn er niet echt verrelde critical bugs voor de volgende verrelde. Er zijn twee dingen die we kunnen doen over hem. Nou, we kunnen ze fixen en make sure they migrate control shift. Wat doe ze doen? Plus, een andere keer. Dus, het is beter nu. Dus we kunnen twee dingen doen, namelijk fixen ze en make sure they migrate to testing. Of we kunnen de verrelde verrelde van testen. Natuurlijk, als we de verrelde verrelde van testen, de verrelde critical bug is niet meer in testen. Dus, het is echt een probleem voor hulp om deze verrelde te fixen. Zodra we veel verrelde verrelde van testen zijn. En naast deze verrelde verrelde critical bugs hebben we op het moment een 1000 verrelde verrelde verrelde verrelde. Mijn vormen zijn voor GCC 4.3, die ik later zal praten. En er is een belangrijk verschil in deze verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verrelde verreldelde verrelde verrelde verrelde verrelde verrelde verrelde verreldelde verrelde verrelde ver Voor een geluid boek wil we de geluid belangen als het boek niet bevindt, maar we willen de boek echt bevindt worden. Dus het is alleen een gooi en niet een blokker voor de geluid. En ze hebben dezelfde onderzoek van de geluid boek, zoals Rc-boeken, dat betekent dat er een 0-day NMU, non-maintainerploed, voor boeken die al vijf dagen gegaan zijn, zonder een maintenance-reactie. Dus wat zijn de geluid boeken? We hebben de eerste die in de file gebleven is, full IP version 6-support. Er waren gisteren 30-boeken open. Dus als iemand interesseert over de 32-boeken, dat zou bevindt zijn. Een andere, heel belangrijk, voor sommige van ons, is de dubbele compilatie-support. Dat betekent dat als je de pakket opbouwt, en je het meteen opbouwt, dus dat je het twee keer met elkaar opbouwt, dan brengt het. Dus er zijn 236 pakketten op de moment, dat brengt als je hem twee keer met elkaar opbouwt. Het betekent dat er iets ergens met een clean target is, zodat sommige files worden bevindt, maar ze zijn niet bevindt, dus ze maken de build-fay. Een andere geluid is dat we niet willen hebben een recommandsrelation in de main, die niet bevindt, in de main zelf. Zoals de politie zegt, dat alle pakketten in de main zijn, moeten ze alleen opregen en recommenden op pakketten die ook in de main zijn. Dus we hebben een systeem die totaal vrij is. Dus dat is wat deze release goal is. Dan hadden we een makkelijke release goal, en we hebben het al gedaan. Dus er zijn nog 0 remaining bugs voor dat. En dan willen we internationaliseren supporten voor alle diep confusing pakketten, dat betekent dat alle pakketten die interact met de user during installation should be able to be translated in all languages by using that conf. Currently there are only 5 bugs left. It's a goal that is already a long time being pursued and so there are only 5 bugs left. So let's try to make sure there are 0 bugs left for the release. Then another one that's already a long time candidate is full large file support. There are only 10 bugs left, so also that one should certainly be manageable for Lenny. And we have still some other ones. We want all Debian change log and Debian control files to be UTF-8. There are only 12 bugs because of course if it's ASCII, it's already UTF-8. So there are only 12 ones left that use other encodings. So also this one should certainly be doable for Lenny. Dan one of the famous ones. We want for a release goal to switch PINSH to dash. And there are currently 96 bugs open. Well, only make it possible to make it to default. At least to make sure it's possible. Slash Lenny. Slashgoals.txt. En basically is there all bugs and text bugs for it, so it should be possible to find out. And if you don't find them, please send us a mail and we'll make sure it's more visible. Which is probably not a bad idea, either way. No, we definitely don't bother with the page. An official page is better. That is no wikipage, for good reasons. We don't want people to, well, add or delete release goals. Anyone can add any wikipage he likes. The official point is whether we step in ork. And say ah, say ah, say ah. But anyway, we should make sure the bugs themselves are more visible through the release pages. At the moment, these categories are certainly visible, but not the actual bugs. So that's something we still have to fix. Dan, one, another one is to make sure that we can use dependency based init systems. So that's mostly about headers in the existing init script. En er are currently 90 bugs open for that one. So that should also be possible to accomplish. Another one is making pure parts clean archive. And this means that we want to make sure that when you install a package from Edge, so from the current stable, you upgrade to Lenny. En you remove the package again. That you are in a clean state. So that there are no files left that shouldn't be left. And that nothing breaks when you do that. So this of course includes that the upgrade goes well. Because if you upgrade and remove, the upgrade should already work well or the remove probably won't work either. Currently there are 73 bugs open, but these are the most obvious failures in that category. If we would file all these bugs, that would mean way too many bugs. We want to focus on these already filed bugs. To make sure the most obvious problems are already solved. And then the one with the most bugs filed currently is for supporting GCC 4.3. 445 bugs. Well, these bugs are really easy to fix. So please people fix them. It's mostly just making sure to include directive in some files. So it should be really easy to make sure that the 445 gets fixed really soon. And then one of the recent added ones support for Python 2.5. Currently we are using Python 2.4. There are only 16 bugs left, so also that one should be fairly easy to accomplish. Just today we decided to add another release goal. Namely the transition from G77 to G4.3. To make sure that we have a more reliable and more recent 4.3 implementation. That should be about 20-25 bugs. To release schedule, probably the most interesting part for many people. Well, in a couple of weeks early March we will announce the very soft freeze. This means that we will urge all maintainers to not upload any stuff that isn't meant for the next release, Lenny. It also means that you really should be thinking about what still should be in the release and what isn't sanely meant to be included in the release. We will also freeze the list of release goals. If you still think there should be some more release goals, please let us know as soon as possible. We will start the BSP marathon, so bug-scrashing parties. Real-life bug-scrashing parties mainly, but also online bug-scrashing parties. To make sure that all the release critical bugs and most of the release goal bugs are fixed within time. Early April, we will then, well hopefully early April, we will freeze the essential toolchain. So the essential toolchain that's mostly the built essential packages. Then a bit later, mid of June, we want to freeze the non-essential toolchain. So that's like DAP Helper CDBS, Quilt, D-Patch packages that are needed for many packages to build, but not for all of them. We also want to freeze all libraries. En dan, of course, we want to have full freeze about one month later, mid of July. And then hopefully release Lanny in September. So how can you help to make sure we can release in September? Fix, of course, the release critical and release goal bugs. Ja, as soon as possible, actually, because if you don't fix them, certainly for the release critical bugs, we will be trying to remove the package from testing. And like I said before, only upload to unstable if it's supposed to be in the next release. If you're not sure it's already released already, please upload to experimental and find out if it's released already before you upload to unstable. Another thing that's quite interesting is to please refrain from uploading if testing migration will be worse if you upload. So if you see that your package is in a bad shape in testing and the current version in unstable is released worse, please let it migrate before you upload another version because every upload can delay the migration. Also try to not upload packages if they are already entangled in another library transition, for instance. Also an interesting thing is if you want to help speed up the migration, it's important to use the correct upload urgency. So if you fix release critical bug and it was a small fix, please don't upload it as normal but rather as medium or maybe even as, yeah, rather upload it as medium. Also for the security team, if you fix security bugs, mostly security fixes are rather small, they are important to our users, so please upload with urgency medium. So they only have to wait for five days and not ten days because in the five more days maybe another package is uploaded which will delay the migration of your package. And probably the most important thing if you upload packages, please test them before you do. Also if you're not a maintainer you can of course help the release. Namely, if you find bugs please report the bugs because we want to have a really high quality release so every bug that is not filed before the release will be trouble after the release. So please, if you find bugs, report them. But try to make sure they are not already filed. So please don't file duplicates report bugs if you find. Another thing you always can do is send patches to bugs because if someone finds a release protocol bug with a patch it's more likely to be uploaded soon than if there is no patch at all. A thing that also everyone can do is participate in bug-squashing parties. Also if you don't have any previous knowledge whatsoever because especially real life bug-squashing parties are there to make sure that new people can help to fix bugs. Of course other things you can do as a non-maintainer are to test upgrades current edge to testing or even help in preparing the release nodes. So if you test upgrades you probably will find issues. Some of them can be fixed before the release. Others are probably just not fixable or not in that time frame and should be mentioned in the release nodes. So if you do find such issues please report them to the release nodes package. Yes? How do we know when a bug-squashing party... Every bug-squashing party should be announced to the WN Devil Announce mailing list. WN Devil Announce is only like 15 posts a month so it should be manageable. We plan on once to organize a bug-squashing party of course he's very welcome to do so. Does it depend on the organizer of course? Real life bug-squashing parties have from like 5 to 50 people attend. Online bug-squashing parties well it really depends. I once organized one where I was kind of the only one. But there are also online bug-squashing parties with hundreds of people so it really depends. And of course everyone helps because everyone is focused on fixing release critical bugs or at least release goal bugs. So are there any more questions or remarks? What about packages that were removed for instance if a maintainer was missing in action and such packages could enter are there any special process except from issuing a reference and waiting for someone to follow? I have an example. PHP Group used to be maintained by someone who is currently missing in action so the package was removed by you and others. Maybe there were other points. This party made security problems in PHP Group. Do you see what happens every time the release team start from security problems up and says no security. It's also a really big pain to update. I know all the problems. Let's say that there is no security issue which is not in the bug tracking system at the moment. Normally if you want to just reintroduce the package in the archive so in unstable you just have to make sure that a dabbling developer uploads it en that FTPmasters accept a new version. What's the deadline for... It all depends because if the FTPmasters don't accept it before the freeze it won't be in landing. Is it removed entirely or just from landing? This one was removed from unstable and testing. So I need a sponsor. Doubt who gets one for that package? The problem is we have a specific case. We have PHP Groupware which is somehow a bit replacing PHP Groupware. The other point is we have been... We have to learn from Lenny and Sarge that it's a wise idea to let the security team meet you against packages because in some cases we release with packages dozens of security holds and we have to fix them every 10-12 days. And after some time it gets quite tiring and especially if a maintainer goes missing and our upstream is not interested in supporting all versions which is, for example, the case for Mozilla and where there was a case for... We can't meet with Mozilla, unfortunately. Let's tell the case for WordPress. For example, WordPress which is not supported anymore in Stable. So we agree that the security team can decide that the package should not in no case enter a stable release because the security team is responsible for giving security support for all packages. So I guess a package like PHP Groupware which would be reintroduced to the archive right now will not get into Lenny basically because the security team will say, okay, we don't know the maintainer. We don't know if he can help with security problems and so we will not be able to do it ourselves and maintainer is not trustworthy so please don't add it to Lenny. In general, if you are not speaking about PHP Groupware but normal packages then, well, you have seen that looks like there will be a freeze for all the packages so anything that is... that can go to testing before the date which really needs to be uploaded 10 days to be applied to the date, usually if there are no bugs in all found can go to testing if uploaded later, not. Of course, with exception to work it seems quite important or so that we will start earlier if there are very good reasons to add something later yeah, might happen later but as later, the more important this needs to be of course, because at some day we want to say well, at least, which means totally 100% frozen yeah, so how it's going to be. But for some packages I don't know if PHP Groupware is in that case but some packages it might be interesting to not have them in the release itself but to have them in backwards We have a PHP Groupware to broken for even that I don't know Sorry There will be more backwards like the backwards my modest opinion are not covering quite a lot of software It depends basically We have different... We have the wallet file for packages which are both hard of debbying and which are fast moving especially Glamaf We have an update of we have an update of Zeezer data means an updated time zone files something like that and then we have this backpots org which is basically for channeled backpots but also in that case like in other packages debbying as somebody whose debbying developer is going to maintain it or nobody is going to maintain it that's how it is in debbying Backpots is not an official part of debbying especially the release team is not responsible for backpots so feel free to ask persons who are responsible who are not here today nobody from backpots team is here but usually every developer should be able to upload and maintain packages feel free to do it and you can be sponsored to backpots and even non-developers can get access at some point so you don't need to be in a new method of access Yes We also want to see if you want to enter a backpots org it should be a death in also Yes Well that's the general rule but for some packages there are exceptions It's nice that people look at what the release team does but it doesn't mean I have a very poor testing we as a release team have a responsibility if something is broken we need to fix it for backpots org ofc they respect whatever they have for the testing but in the end it's the decision of the backpots org they say there is a faster package which they have done more than once Yes That's their decision A strange thing no one asks about Armel So curious So as probably many know Armiel the new ABI version of Arm has been added to the archive There is already testing migration happening for Armiel At the moment we don't want to say that it will be included in Lenny because there might still be issues well there are still some issues but we are hopeful to have Armiel in Lenny I'm bored So it will be next to well probably will be next to the current ARM port We don't intend to support ARM next to Armiel in Lenny plus 1 Just to be clear Armiel is supposed to be the future and ARM is kind of that Yes Just to tell you for the release how do you see the essential to change the free about one month after the such free What are the rules for rule at that time to say ok no we will free to change One important Release physical breaks in the full chain or basically experience from the past We looked at what did we plan what did we do for it which is not exactly the same anyways we said ok what should we learn from that what do you have to do for example if you have a basic toolchain freeze that means that the compiler version is stable which means a few hundred we have to go it's not only about what we can do but what all developers can do because we don't have a freeze date at this date we want to freeze it and we should have less than this number of a C-Bucks because at the end if we release we need to have zero a C-Bucks or at least assume so so that's what happens in between we stop it with release of charge we said we want to freeze all of our packages on this date and 2 days prior to that somebody uploaded a build demon helper named cdbs which just broke 150 packages we are not going to repeat that so that's how we get to it of course we want to learn something from lennie we want to have a difference can you even difference can you and so on fixing the bug in the build champion yes of course if you change for example the new compiler you can easily add 200 a C-Bucks that's not difficult so you might want to prevent that happening to latest schedule well basically we have decided on the schedule and ready and we will talk about the schedule until the final freeze date which will move all other more or less fixed early june or early march which means we can decide in about a week we are happy with the current state and even if there are still C-Bucks we can and will freeze the packages and manually approve the fix for those C-Bucks because the main point of the tool change freeze is to give people the basis to work on we are guaranteeing the exit package which was the archive at that point but it will be a package like that with only some fixes applied similar enough so we don't have rules like we decide now it's a good time, it's just the schedule so that's a fitting date and we do it so no other questions do you want to I didn't talk about it it depends on the portals let me quote a good example a good example is the army L port the portals made sure very most packages built reasonably well they work together, they just work can we go to unstable we are able to unstable we are able to P-Masters we are able to unstable in one week later I think 11.000 army L packages unstable compared to 30.000 of arm packages which I would say they built most of the archive already in that time so now we allow the packages to flow to testing we will give you further but for her, her is nowhere near ready to be used we were probably not even good for any plus one because currently not enough people are interested in doing that in addition of a new architecture a completely new operating system is much more complicated than just adding a new ABI like army L especially as we have said we have said only take army L if you can stop arm afterwards we are dealing with architectures it's not so bad because it makes it easier for us in our mirror size the mirror size doesn't grow in long term because we are dropping the arm packages probably in the next operating system it will be K-free which has seen more work than her in the last few years and has been more ported K-free BSC which is basically a free BSC kernel and no user end but if somebody wants to change it start as a porter for her that's how it if the porters do the job very well it's not that we don't want it it's just the current brand importing looks like that no it's a bit early to to say so no other questions or remarks so i guess the security people will be happy so i guess that's it if you have any if you have any more questions