 It's there on the schedule. This is the processing template, right? Processing templates, yes. So does anybody want you to be on the stage? Do you want to be on this? Anybody? Everybody's looking at you. Sir, there is a whole thing that is in the same beginning. Is this recording? Is it live stream? I don't think it's live stream. There is a live stream. It's recorded, but it's not live stream. It's live stream, fine. I don't think I'm going to press something. I'll stay here. Live stream, let it go. Is this live stream? Yeah. What was the number? Okay, fine. No, it's live stream. Live stream, go. So, basically the form of templates is that the template is just a page on one particular page. Oh, right there! Okay, thank you. Every template is a wiki page on one particular wiki, which cannot be seen from other wikis. So, if a template is used for another wiki, it has to be a copy of a wiki, and translated separately, and then it's forward, and you need to manipulate them separately. And this actually happens a lot. There are hundreds of wikis that have been maintained, and thousands of templates. So, this just works like anything. The hundreds of templates, the hundreds of wikis, turn the application everywhere, and the infobox is one of the effects. For example, it's really not just infobox. It grows really young in infobox. One example that we have is the IPA templates, for example. So, in the English TBA, for a lot of foreign names, for cities and people, there's IPA pronunciation, and the process after the particle cycle. And this is really useful for leaders, for the very few who use the question, no IPA. But the whole of that is that that's the thing, that TBA is in all languages. Now, what will it be as to have an IPA template, which is basically this kind of, actually the same as the English template, but there's several problems with it. As the characters of other templates, the IPA template can have different parameters from the one in English. And what's even worse is that constantly it has so many articles about and it has so many languages with names and so many languages. But even the IPA came out very well-organized in this. And the IPA template in English has actually lots of different templates. It's not just one template, it's not just one template, but at least one template for each language. So there's an IPA template for Chinese, an IPA template for English itself, because an English language means that our character pronounces the word. An IPA template for Russian and so on and so on. So there are actually many dozens of IPA templates in English itself. Why is this a problem? Because before I would say a lot about the language, then I should also translate the name, the pronunciation of the name, but if there is no template for the same name, it's not going to be a template in my opinion. And I mean, from another template, I could use only a single template including all the decisions that I have to make to translate the name. And I have helped many of them to translate it. And the whole point of the conversation was to help people to translate it more easily. And you don't have to make all the decisions along the way. You won't have to look for what to do and so on. So I want to help them to just translate directly and not open lots of browser tabs and look up things and so on. So if I'm translating an article from English and there's no such template in my wiki, it slows me down. I want to avoid that. So IPA is, again, that's just one example out of possible lots of other examples. It's not just infoboxes. And it's not just IPA. It's lots of other things. So all templates are forks. There's no unified way to manage them. There's no unified way to deploy them. Once you've worked them, they just keep developing. There's lots of talent there in the different wikis. Lots of effort, lots of knowledge, which is not really shared. It's more of a community problem. I am not aware of any mailing list or any kind of forum or anything where template maintainers from different languages communicate with each other. I'm not aware of anything like that. I know that there's a lot of... Really, it's essentially software development done in a very weird way. But this is software. And people, for example, develop various tools for linting templates. All of these points are in the user pattern in case you're wondering whether it's written anywhere. There are lots of tools for linting templates. So you may have noticed that the English Wikipedia quite frequently in the citation templates, there are various red warnings, missing parameter or invalid parameter or invalid unicode characters and in this parameter and so on. You may have noticed this in English Wikipedia. There's a huge stuff going behind the scenes written in the wikis index. Some of it is in Lua. And this is only for the English Wikipedia. I'm not aware that this is replicated in any other language. In the English Wikipedia, there is something comparable but completely separate. Done completely separately. Outside of the Hebrew Wikipedia nobody is aware of this. But it's a simple thing. It checks whether the template parameters are correct. It doesn't show it in the template itself. It shows it in the box separately if you enable the particular gadget. Which is, well, it kind of makes sense because if these errors... You can say that these errors are not really useful to readers but they are useful to maintainers so you can enable it as a gadget and then you will see those errors. Different solutions and different wikis... Each of these things could be useful to a lot of other languages. But currently it is only maintained in one language and it's very hard to replicate this to other wikis. Because, again, it's not written in any kind of source repository. It's all just separate wikipages and separate wikis. So, it's a problem for content translation. It's a problem for... It's a lot of wasted effort. A lot of very good effort that is created. A lot of good error-fixing, design improvements and so on. Happened only in one language. Not necessarily English. A lot of curious stuff goes on in Polish. A lot of things that I actually saw. A lot of curious stuff goes on in Russian. A lot of curious stuff goes on in French. The English Wikipedia doesn't enjoy that. The English Wikipedia could enjoy that. Not everything is invented in the English Wikipedia. Finally, one last thing. Is there anybody from wikidata here? I don't think I see anybody. I will mention wikidata anyway. Wikidata is very frequently mentioned as kind of a solution to a lot of template problems. It is indeed a solution for the data in wikidata. The actual parameter bounds. A very common example is when you have an info box about the city then you need the name of the mayor. You can't take the name of the mayor from wikidata. That is a solution. This is not what the session is about. There is a bit of a problem here about how is this taken from wikidata because it is stored in wikidata but when you actually implement this it is done differently in every language. Yet again the same problem. There is a very complicated way in the Russian Wikipedia. It is like layers and layers of different templates which eventually make this very easy for the article editors. As an article editor you can eventually just write the template name in the article without any parameters at all and then it just shows all the data from wikidata and even more than that straight from the Wikipedia without going to wikidata and that is implemented in very intricate template language in the Russian Wikipedia other wikidata languages cannot enjoy this. I am talking really about the implementation of the template itself not about the data not about the parameter values. Wikidata can solve the parameter values but still the way you pull it is different in each language. Now these things could be unified. I am not sure how I am aware of some existing paths, some existing proposals to solve in this problem. So one very popular way which was discussed already several times there was a session about this in the last years Dov Summit presented by Lego who is unfortunately not here today he will hopefully be here tomorrow but I did try to bring Subban team here to speak about this it's called the Shadow Namespaces so this is supposed to be a technology and Subban will I hope give more details this is a technology that is supposed to write the template in one wiki that will act as a repository for other wikis and it will be available in other wikis to reuse. So this as far as I understand resolves the issue of actually having the code of the template available in other languages however by itself it doesn't resolve the problem of localization so an info box for example it needs names for the labels in the info box like the word mayor it has to be translated to other languages Shadow Namespaces by itself doesn't resolve this problem possibly we could pull these names as labels from wikidata or maybe we could create some files with these labels in a second and then get translated using the translated extension or literally a few days ago literally last week Yuri suggested a way to do this which actually already works to use these data files which are used for charts but it actually already works cross wiki so somebody had together a way to write template strings for use in templates as part of these chart files and this actually kind of resolves this resolves this issue but in a really strange way but maybe Stas knows it better are you familiar with this yeah if we can just come here we will be also seen in the video so basically we can store data on commons now data sets and these data sets are available to all wikis and there are not a lot of data types but for the purposes of this talk the data type is relevant which is basically a set of strings one at each language so you can have a data set of labels for the template which would be rendered on each language with their specific labels I'm not sure it currently works with templates out of the box but I'm pretty sure it can be made work I don't know that doesn't already work I'm not familiar with templates to I haven't actually seen it in action I only received the email on Friday I think so it's like the brand new thing but it kind of works already so it works for like graph extension and what else data visualization and so on so there is no reason why it can't work for any other functionality so a practical problem with this that I can immediately think of is ok so maybe we can write the translations for the strings and templates but it will look like wiki pages it will look like text files translating stuff in text files is very very old fashioned we have such nice systems for getting software strings localized mainly the translate extension and that's very great come on in we were just talking about your stuff again it's good stuff so yeah that's what I was just mentioning and please correct me if I'm wrong so the data files for charts right so you wrote last week on wiki.l that the same files can be also used for writing strings that can be used in templates but it will look like text files there is no text file there is pages wiki pages so there is a wiki page of wiki content and there is a wiki page with a json which is data set data set is a json table you can have in the table different types can you maybe come here cool so there is basic types just like json has basic types string, boolean, null if it's a type and number and the localize there is one magical type that was added to this basic json specification it's called localize localize is basically an object it's a key value store where each key is a language code and the value is a regular string which means that you can actually have a if you actually open one of those it's all connected to the projection never mind so if you create a data set a very simple data set with two columns string id column and localize message column then you can have a very simple lua script that you can say lua script here's the data set with these messages here's the actual key for the message, here's the parameters for that message create me a string and that's what lua does this module called tnt it's already english wiki for sure you basically say invoke tnt then you say message msg this is the data set that contains my localizations this is the key this is the actual key of the message I want and these are the parameters that it should substitute and that's it almost no different from a regular lua invocation it's a linear that's the only thing is it linearly goes through the table but that shouldn't be that big of an idea because considering that jason parset and whatever it's non-compare just a second so it's a regular lua combo this just a process you can stay here so this whole session was a bit improvised and by me so I have a proposal at this point so Yuri knows these data files and subu that are cool thank you and subu and team know about shadow namespaces if you could both of you can come here and have a discussion about bringing them together that would be perfect well data sets by definition are one wiki only there's some there has been some questions why don't you also put it on other wikis thing is I feel this is a mistake actually even in a shadow namespace my personal opinion that it's a mistake to allow local overrides because local override implies that there exists a language only specificity that need to override the main master one because in my mind there's no language specificity there's usage specificity in other words you have a data set for most use cases and then there's a fork of that data set that has some other use cases it could be language boundaries but it could be multiple languages for example I don't know there's a major disagreement between Russian and Ukrainian as well as Moldovan as some other former Soviet Union republics there's some disagreement and they want different data sets but those are two data sets that are at the top level and they both should be shareable between multiple wikis I would agree that there's use cases for non-overrideable remote transclusion good yeah so I think RACIS RFC started out as a way to refactor and unify a few different code paths that were doing pretty much the same thing specifically common descriptions global user pages and the thing in media wiki called scary transclusion so Kunal was looking at basically unifying those and cleaning up the code I said pretty much from the when I first wrote scripted that I was that I thought remote transclusion should be explicit and that did find its way into the RFC I think as an open question whether it should be explicit or not so yeah shadow namespaces does not solve all our problems or even it's it only goes far as it's an RFC which sets out a few of the questions that we've been talking about and doesn't actually answer them so that's the state of shadow namespaces and the work that Kunal has done so far has been mostly in the past to make remote transclusion work properly in a modern kind of way he's not really concerned about getting on how it's been a function in terms of local versus remote but just given a remote page please transclude it and make the transclusions work yeah there's two types of remote transclusion you can pass remotely or you can pass locally you can either pull in HTML via the API from the remote wiki or you can pull in wiki text and then pass it locally so I think he has both of those in there pull in HTML you know common subscription pages you pull in HTML could it be an infinite recursion there yeah probably is that why it's scary in a recursion that's not the reason it's scary I think it's scary for its own special reasons yeah when I was designing the poster a question so we were talking about templates sharing templates that seems to be a bit easier but it's very simple so sharing and also on the community wish list was sharing of gadgets so sharing gadgets, sharing lower modules seems like an easier thing than sharing templates which seem like the hardest of all the use cases why do you differentiate lowest groups and templates in my mind they're actually they're similar in terms of they both take a list of parameters and they produce wiki markup yeah they're similar the code doesn't need to be translated the law is usually called it doesn't need to be translated the code doesn't because the strings that it shows they will have to be translated they can unlock and you already showed the translation usually you do have it can load many wiki messages but where do you actually you can use messages and that would actually be good if you create the messages somewhere and you store them somewhere and you give away to translate them but the translate extension will be very easy the only question is where is the repository where you start creating them and that's like creating a data set actually that's the funny thing because the data set is the position for that efficiency aside that's why I brought you together here to discuss this so we probably need a generic way to have localised databases like a localised stream tables and which can be used from both law and templates I guess the difference between law and a template is just the language you're using the reason templates aren't usually localised is because it's not very convenient to localise them you can do it in theory but complicated things are more it's better if you have to do something really complicated like language dependent formatting it's probably better to do that in law depending on how complicated it is but if you look at the TMT the way it's done right now it basically just creates a raw message that's the parameters so none of this complex code is actually even but it doesn't have a translator friendly UI does it? you're right you're reusing an existing feature but so as a result of that it hasn't had a product development where we're sort of considered how we're going to make this user friendly right now in order to edit this dataset you have to click edit source and you go into JSON and you find the ways you need to edit and you edit JSON and this is horrible 10 minutes ago I spoke with senders and maybe someone else who is interested let's work on a good spreadsheet editor for datasets I don't know if you're just hoping to translate that it's not about the translator but this is more than that translator it's just a generic table editor but it's not another I mean the translator but again the translator will strain by itself it's very hard to make these individual cells connect to translate wiki because translate wiki has an IT concept a global IT project not really easily because then whatever you use to generate it based on it's a huge I've already kind of thought about it that's not a simple thing but the point is that we should build a spreadsheet editor and as far as messaging maybe datasets are wrong a wrong way to do it it's just a convenient and the most important thing it's already there as opposed to all the other theoretical ones that are not there did someone want to say something anyone want to jump in who wants to work on a dataset that idea of how do you control the repository with an open base that needs a user to have something like that as a user you would be something like that a lot of people are only working on their one community and they're they don't care what's going on elsewhere yes I agree don't you think we're going to lose some people we lose some stuff elsewhere I don't know if you may not be aware that for many reasons there's a huge now people are going to think about digital data because they don't understand how digital data works it's outside of what they do it's not working like a digital data so there are regimes that do it so I see this as a very cool thing and people have the sense to learn from digital data experience but at the moment we don't see how it will be possible to compare people to our digital system because they don't care about the specifications they don't understand the stuff on their computer and secondly there is also that risk of losing people with skills to create campaigns but who are used to create campaigns using digital data to test what's the new system we already have that tool with digital I remember we used to have a lot of people able to create data to create new campaigns and at the moment we are only 5 or 6 people able to create new campaigns to maintain new campaigns so I'm afraid we are going to lose people if we are going to have less people actually I'm also saying this because as a community I have to be very pessimistic to be able to see the risks of that so it's an excellent question and I actually I spoke about this briefly and I already wrote this in my initial notes there you can see it in the etherpad yes there are a lot of packages and sharing templates between projects but there will be this problem which you describe several problems one, yes, there is a lot of skill already a lot of talent, a lot of effort a lot of experience in different wikis and most of it is per wiki and it hardly ever it is hardly ever shared between wikis so if such a thing is done there will have to be some kind of community what we call community engagement I don't like this name very much but that's how we call it when I say I don't mean the foundation I mean the developer community the media wiki developer community and the templates developers communities which are many islands in different wikis we will all have to talk about this together because the skill is there the on wiki experience is there and a lot of wiki syntax writing experience is there and it's indeed would be a very bad thing to just give up on it and lose it not just in a human way but in actual technical way we really should not lose it that's why we kind of have to bring this together and yes there is also the question of the feeling of losing independence because every project projects may have different opinions about politics or whatever projects may have different opinions about how templates should be designed the french wikipedia has a beautiful completely different design for the info box for cities it's beautiful if you haven't seen it on french wikipedia they could look something yeah it's beautiful and I have absolutely no reason to force a different design on the french wikipedia I'll be happy if they can just keep using it that's fine but why call it french why not call it template info box style 2 and then whoever wants to use whichever wikis want to use style 2 that's also possible I don't want to expropriate any independence basically copy it sideways yeah unless you really anal about the naming of the template and there's a naming conflict right there okay now but you know this one is much easier to deal with I think than the actual content so I think you understand that it's like you're talking about more of a total repository of templates and I think your concern is mostly that sharing is not going to happen because you're saying templates can't be shared apps have to think they need to share are you saying sharing is not going to be very important is that what I said yes and no it's more about you know that's all the changes that I deal with my wiki it will not be possible to do that yes I mean the reason what you're saying is that sharing is not going to be very important even though it seems like a development for templates sharing is part of the process of translation how about not simple very complex but very complex solution where the template namespace actually for everyone I know we can transcribe template namespaces too all inside one virtual namespace when you transcribe insert template blah that works but you can also insert template English blah basically put all the templates from everywhere into one repository but make on each local wiki appear at least the one without the prefix is the local with the prefix it's on the other wiki maybe for the message or a witness and also have a local place where I can keep everything every time you keep it but that's the thing it's not going to be you will not have a sense that it's living outside of your wiki because you will just have it all living in a still template namespace but you will also be able to use other templates from other languages you will just have some prefix kind of like common where people will know that this is designed for everyone and then by extension by continuing from there you can kind of like say but you can also switch to some new domain which will show you all these templates as equal as basically equal as opposed to one more preferred than the other so we will have my field we will have a context problem because there are templates that refer to other templates and when they refer to other templates they don't know that you will have this that you went to other wiki and took it from there they refer only by name so would that name mean the same wiki or would that name mean the same template on local wiki and each case might have their use cases and it gets and some templates also use like local or they could use messages and so on and that would also kind of Tim has this been proposed? there are some complications basically one virtual namespace putting all templates in one virtual gigantic repository where they is it too complicated to implement or is it there will be issues in our so I don't see exactly though it will be complicated but it will be quite big complicated I think there is a technical issue that is social issues but what I am hearing I mean it feels like conceptually we are talking about mostly a third in principle but no shadow I feel shadowing is a shadowing is a good question because shadowing is proposing a standard wiki in which you put all templates and your objection primarily is shadowing or not basically I am very much against the idea that wikis should be separated based on a language because I think language is a very artificial boundary that was introduced purely for technical reasons initially and I mean we could not handle multiple languages very well initially even now we still have and have and that kind of what is your name Benoit Benoit you were saying about losing people who know template this is a problem for all of the wikis and unless we bring all the talent into one pile so that people can cross benefit from each other it is a dead end you cannot have 300 or actually what 900 900 developers and template developers doing as good of a job basically replicating each other and even copying is not it would be more interesting to have 300 developers working together instead of doing some job but people have 600 people asking them but as always I think it seems I mean what I have been doing and this is my personal know-how I just created all my templates mostly graphs because they are so complicated so few people actually understand what the hell is going on inside I always create them on media wiki I put a big disclaimer at the very top do not change this unless it's on media wiki.org and then I just copy it as is to all the other wikis yes that's exactly the part that they want to skip but copy pasting is actually very quick and I can automate it with a bot it's very quick if it's one template but that's the thing it has three templates and a lot more so I wanted to do another wiki and now I need to copy and paste that is the problem that's the fundamental problem but here's the thing, what I proposed a while ago is the do you remember the RFC on the budget copying on the budget thing basically multilingual templates on the budget something like that the idea was very simple you create a template on media wiki for example you put the big disclaimer and then you mark it as an auto copyable template which means a bot just sits and waits for the main one and then it looks at the wiki data all the translations of the template and automatically copies it whenever it changes it's very simple it just basically copies the template because of hacks like TNT the translation system you can make the template completely language agnostic and that's it you don't need to do much more if a community decides to stop we're actually going to fork the community places some sort of a template at the top or even just starts modifying it it has been marked it has diverged right you can no longer copy it the bot can do some sort of a notification or whatever then we can figure this out but the point is that just have a bot automatically copies a set of templates that has been marked with a special magical it is a hack but the thing is in media wiki everything is a hack I mean who does not think so please raise your hand my point exactly it proves the point you're right, not everything my code is not because I brought it here because he said that you object to me saying that TNT is a hack so don't object anymore I do not object to it being a hack only on a condition that everything else is also called a hack I want to go back to this it looks like more and more templates seem to have our properties it seems like rendering the information is in visual editor how it should be rendered to be serialized and how it is about whether it can be copied or not whether it is local or not it almost seems like an abstract of our business I don't know are you talking about template metadata template that extension has the localization also for your support template that extension needs localization of each parameter in a human readable form that support visual editor so that's another localization for each right, but that localization is stored inside the template which means you can have to copy the whole thing over no, it's not in separate underneath or in the documentation page yeah, it's in the documentation page again, it's not an essential locator just a question when you're on the time being on templates are you thinking about info boxes as the main motivator for it or not really actually, I would love to get all these nice little cute notification boxes all these simple things out because those are very very simple and yet they are needed everywhere so even getting that little box and more or less standardize on it would be awesome and by the way just a personal feature request for Tim I think, can we please have a kind of like UNC style page title in other words, no matter where you are you can always refer to a certain title by a certain string because right now, we cannot say for example, if you write commons, colon, blah blah blah that's a title on commons but if you do that same string on commons itself, it goes wait there's a commons namespace, no we do not have it it's part of the title which means it's not a global it's global except the one where it's targeting, so it would be very useful and I keep hitting this time and time and time again if we would have introduced this concept of a globally resolvable title and I'm sure a lot of other wikis would be very, I mean other outside sources would be very happy with because then you don't really it doesn't matter where you are locally Context independent title there we go well, wikidata is that, but first of all wikidata does not contain everything it does not contain like doc pages for example or whatever doc does not contain and a lot of pages even in a normal namespace they might not have a data page it's not a globally unique I didn't say it has to be global well, I mean it is globally unique but basically a globally resolvable title you can have multiple of these resolved to the same value that's not a problem just as long as if you put a prefix commons for example it would work everywhere more, I think except a certain I would I already have a hack inside the TNT that says if the local namespace is if the local wiki is called commons media wiki do not insert commons column in front of the link to the data set because the data set at the bottom of each graph I should have a link to the data source which means that if I have that graph on commons itself that link is broken what is I think it sounds to me which part I'm just having a global repository at least alongside local repository templates but at the same time it's important to look at possible long term outcomes in terms of tangency for marginalization and not just technically so like it's as if over the long term there could be a danger of conceptual marginalization out of technical experience for example I'm sure on different wikis languages there's different concepts for these headers that sometimes appear on the page of the templates this page live citations this page self-promoting but policies differ to a certain degree on the wikis and if all of these header tags that are used to indicate problems with articles were centralized and it was the same concept that would tend to like tend to gravitate towards using that and then using only the same concepts and that could for the long run take away from the diversity in terms of approach in terms of policy etc well I mean just because Swahili wiki introduces its own header box and starts using it very actively does not mean everyone else will immediately pick it up and start using it or some community can say hell no we're not using this and we're blocking it wait I'm talking about media viewer for the long term it's much easier to take the ready made thing instead of saying hell no we don't want this but copying a thing there's a tendency for diversity there's a tendency is to maintain that first rather than just gravitate towards the lowest part of the level think of it this way I would think of it this way that the barrier of entry into a new template usage is so tiny it's a copy paste select copy paste that's it you're done unless it uses some different sub templates but even in that but usually for something simple it's that if a user in one wiki who came probably from another wiki and likes it brings it over I mean there is a cross pollination between wikis and that's how we have existed but also there is all this other community in that wiki will say look at that template say no we don't want to use it so I feel there's enough reactionary forces or different forces forces in different direction that will prevent that especially because of the language barriers which are there currently I notice that there is in a lot of smaller languages I look at a lot of wikis in a lot of languages big and small all the time and there is a phenomenon in many wikis especially the smaller ones where there are very few editors very few people actually editing the wiki like less than 5 and a lot of what they do is just adapting templates from English or Russian Wikipedia they don't actually write content they actually write a lot in the template namespace because they think that okay fine we got our own domain language called wikipedia.org now is the time that we can start adapting templates and it takes them literally years until they actually start writing articles because no no we are busy with templates they are really proud we made this achievement took us half a year we adapted the whole info box for cities true stories happened in many wikis and I'm not talking about the really really small ones I'm talking about those that have like a small number of editors, 5 editors, 10 editors we want that to happen I prefer them to be lazy and take the readymade template for cities from some other wiki and write some actual content and when they are mature enough their own info boxes fine let them divert from the usual whatever I'm happy for that but I really want these people to concentrate not on this repetitive technical work but to write actual prose that's one thing that I really want to resolve with this simple repository idea yeah that's a possibility if you want to use the readymade thing translate the labels to your language but translating the labels of a template to your language there are very very complicated info boxes where you have like 100 labels translating 100 labels takes a couple of hours no more than a couple of hours I'm talking about cases where it takes weeks or months to translate a template completely you know how lovely it is to write a condition in wiki syntax a condition it takes a lot of time to do that and I want small languages of India spoken by 80 million people but you have people there busy translating templates from English instead of actually writing in that language I think it is a common problem and that compared to the potential homogenization of I think it's not known nearly at the level the problem is humans we do not have enough to cover basically we do everything on every possible wiki and that's no brainer oh and by the way about the global unique you can do column column wiki name by Michael I was going to say I was saying before that templates are the hardest thing to have in your local repository I guess which is probably not true for all templates especially the templates that are just simple wrappers around a little module I think a lot of the infobox templates are just a simple wrap up which maps out there's one other thing I wanted to make oh sorry no just one thing that we don't actually have to adopt existing templates into the system surprisingly in my opinion we can actually given a new repository it's a chance to actually start a new and create a new set of templates that people will slow it up because most of the time templates by the nature of template language incrementally become so complex that no one touches them so it's probably a better idea to offer this opportunity to create a new repository and say look you can continue using your existing one or there is this global thing that already has localization that already has all the other shareability goodies and you can start there you can start moving things over and I think the community will actually embrace this idea to restructure and refactor and coordinate between multiple languages to create a new set of templates which will be independent of the wikia local templates maybe because again as I said look at the code at the programming very very often we just take a very complicated stinky thing and just dump it and I think we have this opportunity here because it makes sense there is a clear benefit and at the same time we can make it much simpler because we don't have to actually think about migration problem because we can say look there is this new thing yes it means that for the next 10-20 years we might have two of them one shared that has a very clear benefit to everyone and there is a local wiki specific and I think community will engage it because of the instant shareability and translatability some templates definitely made to be shared there is the famous authority control template is anybody familiar with that authority control is anybody familiar with that so that's like that's a template which is very generic it's barely even seen in the pages but it's for connecting to data sources in libraries and wiki data it was supposed to be a global template from the very start instead what actually happening is that people like what's his name Andy something the crazy British guy that picks on the wing I think everyone knows that person yeah he is huge in OSM Andy made him so now he's going from yeah so he wrote books about the rock band Pink Floyd he's great anyway he goes from wiki to wiki and he asks the different wiki communities please create this template you can copy it from the English Wikipedia it's just a bunch of blog code and some templates this thing and you also need to translate this and you also need to do something wiki data and it's really useful for certain things for certain biographical articles and so on but he actually has to go from wiki to wiki and engage with communities and when something changes in that template it has to be changed everywhere and this was supposed to be global from the very start that's why I used copy pastels but it's not very efficient you can do it with a blog at that moment but as I said we have an opportunity here Tim was actually going to work on it this evening he told me no so the question is the whole point of the session is that I'm trying to bring several things together so is Shadow main spaces is that a technology that can make such a repository possible or is there something different shadow main spaces is I think it's starting to deviate whether it's shadowing or not it's still what it's proposing conceptually if you want to know about this it's not a solution and the debate is about whether it's shadow automatically or it has to be exclusive I think that's a very I mean I think what you want to do is go with a positive way and the debate is whether you do shadow in the front that can be a surprise but I think what you want to do is go with a positive way because from the media point of view there's no such thing what you have is just across transporters that are taking questions whether they're kind of so isn't the template complete I just feel that like here we can break away from it and if we try to mix the old and the new it will create more problems like if we introduce global templates I don't think we should mix them in together with the local templates but I think if the shadow makes this with the public forget the shadow it's not a name space it's not a shadow it's a shadow to this and that I'm all for and in that case maybe we'll just use a global syntax for it some might be busy what the hard part of this is we've already got scary transfusion we've said that and what that gives you is you can just write double brace and then the interweaving prefix and then the page title and then you get that page from the other wiki for amateurism probably not but I don't think just making that work is the hardest part I think the hardest part is probably a new wiki to store all the stuff on it which brings with it and then you need some sort of community to do this stuff that's probably not very hard I think people will come together very instantly there's all these product management questions which we've been talking about do you have shadowing how do you do localization once those answer I think the implementation well I think the question how do you do localization in translate extension one way or another I hear why I think translate extension is actually not that scalable translate wiki on the other hand is the out and target translate extension is I mean as far as I know there's very very few people who actually figured out those comments very few people know it understand it that thing is non scalable you're talking about the same extension so the D things are for page translation and translate wiki is the same extension but for messages it has the same user interface for those but it's the same extension it's just somehow interfacing with that it might need to be different so I feel that templates should also be message based rather than content based because there's very few templates that have pages of stuff most templates have just like tiny strings and those should be translatable as translate wiki and much easier for translators to translate someone said was that seven minutes wait you work in translation you know about it why are you quiet you want to come talk here about it there's this thing called translation yeah translation yeah so okay so let's just think about it we make a new wiki and then what and then we make enabled scary transfusion wikipedia and then they can pull templates from this common wiki and then we get messages created like midi wiki messages and then we install the translate extension on that common wiki and then we get these messages translated for the common on that common wiki something like that but that will very quickly grow to many many thousands of messages which is the translate extension can scale to this we'll just probably have to rebuild the project selector like which template are you translating right now this kind of in terms of the translate extension I think it sounds doable Nicholas will correct me if I'm wrong we'll require some work but generally doable in terms of the translation itself is it doable cash in it's such a terrible word which blocks all the technical projects we have the organization we're already doing it so yeah you know you say I said translate extension you said well translate extension has these five problems but it's better to have everything in the translate extension to have everyone use the translate extension so that all the translation clients can benefit from that but rather than to have everyone embed their own solution which makes it sound different I agree I feel like we should target in this translation extension usage on this wiki we should really target messages rather than blocks of text maybe even disable blocks of text support so that it's very important to do the word messages this is not really what's the difference between a message and a block I didn't try I mean you don't want to I feel like there's common things that are wrong common C and 5 whatever maybe there's different sessions but the important point is like the proposal to basically build a new wiki to split ways with the existing templates let's not try to support the existing stuff let's try to have a repository which from the beginning will be global only have but sorry but I can reason that I mean how can you prevent somebody I mean that's more of a policy question but if you create a wiki for templates you cannot really stop it that's a policy question we cannot kind of make a technical it might not work if I mean just copy you have to copy the whole tree and the moment you copy it becomes instantly available to all the wiki I'm not sure I'm saying how do you prevent that technique if you want to prevent that technique that's the policy all you can do technically is create a global wiki somewhere inside of it should not be referencing a lot of templates it should only like the one template right you have to add the templates in the global you can only use the ones in global that's very simple, straightforward so what are the next steps maybe how many minutes in real life like 5 minutes anybody can look at the etherpad I see that some people I actually I actually did some most of the most of the notes they can hear and some people helped me, thank you wherever you are, I cannot see you I started writing some action items so I wrote two action items one would be to create a cross wiki discussion for template maintainers there isn't one right now another is document the current bots that copy useful templates so Stas and Yuri spoke about this about Sparkle and about TNT sometimes I propose the bots that copy it that's the thing that should be probably documented somewhere right that's an interesting idea what if we had community first like we always talk about how we start the wiki we start the wiki which might eventually have global templates whatever else we want to have another world and a lot of people put their templates there and copy bots there's a copy of it going on and basically just promise you do all this work and maybe someday we might enable it because developers feel bot by community problems and the community feel bot by development problems but if the community setting up a community doesn't depend on avoid the sort of GNA problem and set up a wiki and then yeah we can see what comes and then she's the one because they're starting it right then it's going to be over yeah but it looks like I think my summary is it looks like what Kunal is working towards I think if you kind of address the debate about shadowing he's working technically kind of as a technical director we're going to cross the gate and there are a couple of technical questions I think what images are there whether it's explicit or implicit whether there's shadowing around I think those are the questions that maybe somebody looks at but in this case it's clearly it would be an HTML do you agree with that so because you don't want to get transferred because otherwise why would you get the meal because I want to get that but you may get all the slides from there what I'm just talking maybe you want all the slides from the resources from the resources from the wiki and place slides for example what I'm doing is place slides I mean there's a lot of people can place slides where does the slides for can place slides where do they go I haven't seen there can place slides I mean line slides can place slides our time is up cool thanks a lot it was a great start thank you how many people have done thank you 10 to 90 that's the point 20 to 90 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50 10 to 50