 Yeah, so what is this guy? And this is also a lot of possible training. True, we also have interests that we have to separate from these branches and also, of course, it's possible that interest can be directed. So it's not really directed. And as a workflow for development, so she can also be better. The storm must be high. If there are broken branches, they're going to be compromised and more dangerous are attracted to their own content. So also, of course, I prefer to accept the training system that is used to look at augmenting, compiling, recording, and track-sensing. In some cases, also augmenting the training. Yeah, of course. So I said, I don't think it's necessary because it tests platforms that have a very high amount of access to. Typically, people are also going to get a good start at recording in the next couple of windows and there are a few men. So our infrastructure will be the same. All those platforms are typically also going to be used to have an infrastructure much faster than most of the development sector in the central main building environment. And also, it's the advantage of being a uniform with infrastructure. So the developer has an open compiler and the village works for them. The developer doesn't need to also work with a different person that has a new approach when they compile the content. So if you're wanting to just treat it as an error, then it's not really fair. So having a standard system that is going to be on will show at least it's a normal stage where the infrastructure is created and everything is adjusted to the side effects. Because if you want to check out the model that they do, you don't have to have a good break and you just have to write it down or just have to find some other person to do something that would make it really good. Yeah, what you do use in the Neural Synology is a tool for it. And what we are using in the character is our patcher-view system. This is going to be a patcher-style submitting tool for the code to be used by the human, but also by the Jenkins SysDial infrastructure. There's a different phase that is typically used by the USB to set up the other components and the commands. But the commands that it uses may also use a command line and the things that it will do for the dynamic engine. It basically will grab the results that are executed by Jenkins. Jenkins is a system that basically will boost, requires a variant for what the changes are available when it's in the checked, and then has a schedule and system that just boosts the patchers to the different build nodes. And then, after that, we will try to set up the results, submit the script that has a load to the character. There's a plus or minus one load on the electrified components and potentially integrated in our branches via the testing electrified load and a plus two on the code for the code. The code was given out by the Jenkins. In addition to the Jenkins, it's got a lot of inner boxes and those don't go to the pictures but are built from the branch after everything that's been integrated into the branch. This can catch issues that are not to be seen by the patronization but might be because of the problems that have appeared on the patronization. And those inner boxes are also the machines that provide the interior lighting that is procurated to download and test. But what are the drawbacks or intentions in having such a system? This basically points all down to how long do you to wait until you have your results? And the B-Office is an upshare project and of course it takes a long time to build and some platforms are possible for others. For example, you're on Windows you can use cc-cache to accelerate your build and it will also come to the office actually all the apps are free from the standard optimization so we basically have our products made of our B-Office. Windows will also take typically 58 minutes and it's a similar build or it takes 2 hours in parallel and for Windows it takes more than 20 minutes and of course also for Mac you don't have to search for our hardware landscape so we have the slow mix and it's not a good platform that takes up to 3 or 4 hours so you don't forget if there are lots of changes in your build and you don't choose not to engage with hits from the cc-cache. And this is only a one-time equation of course the hardware is relatively despite having quite a reasonable amount of build costs for example F-24 which is making the F-4 Windows in one additional server office sponsored by PiPark that is also doing some things specifically in the biosec-cache. We have a bunch of products of which we have pros, fresh face guide ones and the other ones and of course we have a couple of linear machines but a lot of them are changing certifications like for example fresh pest ones or in-cuff coverage or something like that So the hardware of course is probably adjusted by the hands of thousands of machines that you have and you can just be able to coordinate those like which is also caused by people mostly in one time zone so there's a bit of a community distributed throughout the day so the spike in the European power supply to European light is going to be the key in the benefits but the biggest problem we have far of being random favourites that are not erected are related to a bench that was tested but to pass the problem was the vagueness of the test and the spoil so it's the poor development of the bench, nothing wrong but the train said no I don't approve the bench I give a minus on what we can integrate it and then of course the first part was about the touchback inside of an armored car that's exactly the problem in the test phase this is sometimes subordinate so what has to do with the situation first we try to touch the stones as they are called in Japanese and the touchstone is basically a type of stone the only touchstone that passes the other platforms this then of course is an artificial blade that's one first that passes the first then it can move through the slow system so it might take much longer to get it where you can find it of course this typically works well because the limits ones are usually used by developers developers typically also buy the products or they own some example that is an artificial blade so typically we come to see it then they break some of the limits that are usually also made because these are the platforms that don't have any access to them so it helped to improve the situation that we actually really harshly used but in the end it wasn't because it didn't really work for us then of course we had a simple thing we had a timeout for when it was stuck in a test for example but initially it was just about Jenkins it was an absolute timeout so we managed to save it for the slowest blade machine so the timeout was more than 4 hours timeout so it makes it take 3 hours for a regular blade and of course even if the track is a bit bigger it gets stuck it certainly needs to wait for the 3 hours timeout to be kept and maybe moved from another blade but this now has a change to basically to a test but despite looking at the mid-out there's no mid-out for 4 hours timeout for slow tests the timeout is out because there are no mid-out machine that's still a catch without the machine which was a simple thing that helped but of course one important service only helps in university what we have here is from the matrix project that was used before to the multi-drop to the matrix project that's been a way to trigger Blitz and Jenkins and here we will be making use of the refrigeration which is a moment for things that we have in the operating system and we have 10 versions I can install or 10 versions of the compiler or a couple of them but we are the new growth project for the test product so there's GCC-ATAM and MEG so we can really much use of it in the operating system and have some more functions for example this is the view you get for the project in this case also it's just a category and there's something to worry about but if you were in the operating system and wanted to see the model you would have to use first the configuration name and then we could have the configurations which was the mobile page so the client job project looks like this so we have more detailed information right at the front the image tag as well to also have a battery to hold in the dock in this case it's also a few bits so nothing to worry about and the hot job that's the biggest feature while we were in the operating system is the ability to resume that and resume means resume the one second thing so you see here that's also a trick option and this is also a matrix project this one is just really the only change which is more than the repair on all platforms even for the day so in some cases we have a patch to take windows which we won't use we will build the windows but this of course helps to the other the Chinese business freefall to the other customers and this is a change that I'm not sure of this creation but of course we need someone to actually look at the results and see what can be together since this is another automatic standard case and of course it's about the machine is about faster another thing that I introduced was a priority faster the machine is a higher priority and I prefer to be there who wants to take 2 or 3 hours more of it this is a special help for Mac the difference between a faster and faster process is the most important but windows has a different performance level and for me it's also a faster and for Mac it really makes a difference also I added a fast tracking priority for these branches and of course this is a more complicated for me it's an easy one because I don't have an interest so those are technically intercom for the good stuff from master machine and this also helps a bit to play back a good song in half an hour another thing a system that both do the work and perhaps while knowing the failure states and the preferences that are always the precise and the regular discussions and you can add your own management and it helps just to get the answer and just not to get the reason why it's not a big deal for example I think you can give it a some kind of type some kind of development particularly in Oslo where you can find a good distribution of attention not only about intercom but also both for itself and with this basically it's possible to act automatically twice since now and for example by the failure process what just was the connection to Jenkins and the boot for this case we can just start the alternative in the end not bother the developer for Jenkins and the boot for the invention and right now this is for the case where what was the connection this is for all systems and other two cases that are specific to Windows we have cases where maybe just cracks out with it instead of standing out it's basically a part where we can also to be asked to adjust the this happens to just start the boot if you try also some of that with related related to this so and this is of course also related to the boot or the cache this is mostly related to this is the problem we can have a lot of for example this is the results we see in those will take more than just no actually and also into how we can use our heart there and some more Windows quite much to the large face in the end in the start for instance where lots of threads just a single part and also quite a proportion so we have a lot to have approach to this thing and this was some of the parts some of the parts this was but thanks to this system we can just use the default to this and this is used on Windows and not only for the two different kinds there is a GCC and a clang that can run simultaneously but not GCC there is a lot to clang with because I didn't want to this space just to work with the cache it's better for the past to stay the same also because of our regular processes and sometimes on Windows checking the sign packs to actually do approach and they will have a process and make a subsequent save so we have a single hand that just will sign up so it's in the process and we just have to work it and one of those for most cases it's just using some bit of radiation to keep it working but this also to check out because it's still working because this is the final and of course for example for GCC our default because it's all built with that the status of all the consumers and this across the world we want to try once more to delete all of this so our default just checking before working on showing the overall status not much for us so for people who are actually planning to go for a trip with drugs also for those people and this is basically we send all of what still can be done with two friends this should by the hand be processing fast for example it should be for the next one we can make it possible to do the same so both can also just trace them and it will be just a part of the amount of the data so in this case the system recognises the sorry itself and to take it offline so all of these are actually just working more so that you that is waiting for the process that doesn't be able to product it basically by the end and then just taking all of them and doing some the same thing and it's very important for all in the long run so also we have a problem that can be worked on a lot of countries not only but it's because this process makes all the products and if one of those patches has a bit breaks then it's also very likely that all further products that have this one will parent also but now there's no system that will detect that this is a part of a series of patches that have one that fails it's a long failure so to try to do those this basically is a way to get the results from outside the Q4 other than that so this is something I would like to do and assign the patches here so all of that work is just getting huge for the other way the waiting line or just the stop processing to remain as serious all together and then the patches fix to not why it's possible to have the bit failure on the other side which I'll show you in the video bit result in the way Jenkins can actually search for what is necessary to want the past Jenkins for asking all the inputs where there's a compilation and you you can do it in the businesses that are outside then go and choose the failure result and as a comment it is included but this is a comfortable to use and the only work for this is to search for Jenkins for Jenkins for Jenkins for Jenkins is actually why this is a head forward I want to add this to the final and the error shot part for this time so I can't do this because it doesn't tell when the pitch is hard for the code so again it's a bit of a pitch out but the real code is based on the master branch of the command vehicle then you have a fresh field number and it will go basic so this makes it hard for the first time having a search which some statistics Jenkins won't make it this easy and of course it wasn't really expected Jenkins would be beautiful for the first time it wasn't in the business services to have the same sign on for the first time it makes more sense to have all the committers are complete so you can see when we have Jenkins and Simon and Jenkins and they can do this and it's an object that can be a pooch to receive this and of course you have to also know that this is not an infill but this is a vicious pattern you have to work on the trial system that controls the thinking and try to receive entry in the main order of the pitch but this is not going to happen it doesn't do anything if you actually want to receive it you have to go to the sideline to do that so this is some confusion so I think I can try to write it back to the master what we've got