 My name is Christian Neumeyer. I'm employed by the FBF as a Denise engineer and also taking care of the infrastructure, especially the tools that are needed, especially to work on the ERA office, that includes the tools for the translation work that is done by our volunteers. I'm mostly known by my FBFC handle to SCOF, but you are different from the rest of them, you can call it the office, you can only hate you, I can't remember your name, because I can really relate to this situation because I'm very bad with names, so sorry if I get your name wrong or forget it immediately in the next day, so it's nothing against you, it's just my problem of remembering names. So, and you might know that we are using Utl currently as our main tool to do the translations, but unfortunately, Utl development has stalled and people are now looking for alternatives to use instead, and so he did all the great work and we refer to what Utl are out there, which projects are using them and whether they suit our needs, and she came to the conclusion that Weblight would be the best candidate to use and perform with. So the main points that make a Weblight our pick of choice is that it basically was created as a drop-in replacement for Utl itself, so all our must-haves are already available, so we want to allow people to do offline translations, we want a tool that is actively maintained, that is not just used by us, and we want a tool that is comfortable with it, and not used in any future. And one of the advantages is also disadvantage, like for example, the Weblight is a way of branches and is centered about repositories which should make getting a continuous flow of translations into the code is much easier, so the idea of Weblight is that translators could change the Web interface and Weblight itself pushes it to the repository with some CI system or other that will be the final thing and create a whole closed loop where human interactions needed to do that translation into the code. And you have already said it is actively maintained, it is offered as a hosted version, as a paid version, and because of the feature to have paid translation, we are also tracked by Weblight, we take the benefit of having nice statistics that can be used by the ANC to assess the community environment, how much did that person translate, how re-engagement is, and it is nice to have that we get along with it. And yeah, as mentioned, it is used by other projects already, and Fedora is also looking for a re-placement for their code, which is Sonata, I think, and it is also the end of life basically, so they are also considering switching to Weblight, which also would be a good chance to combine our efforts to make it work for a larger scale project or to save most of the translation. Now, the biggest changes or burdens that we face are related to the fundamental design of Weblight. Weblight basically assumes you have many small projects and those projects have only one or two or maybe three files to translate, which is of course not the case for over a hundred files in a single project. And put on the other hand, didn't care about any structure whatsoever, just looked at what files I did, directories, and created the stats based on this. And this results in a confusing or over-helming display in the Web interface that makes it hard to judge how far is the progress through it, because every file and component is treated the same, no matter whether it only has one string or whether it has a thousand strings, and you have files off-loaded in our system. For example, you have files to list new dictionaries, only have one string. Typically, this is the translation dictionary for language and so on. And then we have others that have thousands of strings, but those are displayed as hundred percent. And we can of course do some workarounds, and this is, we have a short-term solution, for example, for the Overview, we can create the component lists. I'm not sure whether you can see this. So this is, for example, how the Web plate displays the head project. So every single PO file we have is its own component, and before display, only this is percentages, but only if you hover over it, you see we were down. And for example, these two are pretty equal in the string count, but there might be another one that is completely out of like, for example, 20,000 strings, but it is the same, it's all 100%. So this might be a little hard to get the information out of where to start with and put on the other hand was using, or is using a directory-based approach. So it's just, it collects everything that is in this folder into one step. And this can, the further you go, the more you take it, the more split up it gets, and but you have just this single Overview page compared to that in multiple instances on their own website. One way to solve this initial Overview, the problem is by creating component lists. So this basically lets you specify an aggregation-based component. And I created a couple of those for the head project, where I created those lists for the main modules. So this gives them a better Overview, I think, of at least the main components. So, for example, we have almost more than half of the hulk section translated. Oh, this is not all languages, it's not a specific language. But you get the idea, you know that if you are alone in one module, this is not translated. Or you don't have to go through each file and see how many skills are in there. But of course, if you then open this component module, you still get all the individual files and there's no cascading system where you still need to get an aggregation of the hulk. But at least I think this is a little hard, but of course, the idea is really neat. And also, one other easy fit to make it more apparent what needs to be done, this is changed from the percentage display to the string count, so at least you can solve by a number of strings that are left in the trade, instead of having a percentage. And also, you can make the Overview the list of strings much more bearable, at least for languages that have a big translation and completion rate already. But just as they've been tied to completed translations and the options, and then you don't have to deal with multiple pages of BIOS truck to get to the ones that still need some. Also, it also has some more advanced dashboard features where you can set the default huge strings with the suggestions that you need to review or other things like that. Another issue is, because the system is based on the components, there is no built-in way in the plate download multiple files for offline translations at once. And this, of course, affects some projects more than others, some for only browsers, some prefer to only work offline. But as an easy short-term workaround, you can use the repository and get the files from there. But, of course, this requires some coordination in the local team who are going to duplicate work into this upload. And one feature of the mapping that is not yet enabled is the automatic pushing of strings to the repository. So, again, this is something that we need to develop right now, but later once the automatic push is enabled, it should be less of an issue. One of the features of what it has is the propagation of translations within a project with two different branches. But now when you're starting a language and you have the stable branch, the stable branch, and the fresh branch, the master branch, so three strings to translate. And put it, you would have to translate a master, for example, and then go to the 6D project, go to the string, and then copy the string from translation normally and do the same for the 6D to 2D branch. But they would have a feature that would automatically translate the string to the world branches. But we cannot use this feature right now because this would require world branches to be listed under the same project. So you would get triple the amount from the whole component entries. And I don't think this is something we should go for. Rather, I would use the same name as we have in Google now, which means creating a deeper, obviously you are in a master project, and you are obviously in a 6D project, and so on. And it's that the aim is to modify the weblate to allow for the propagation to work across project boundaries. But this is probably a lot of change because it is fundamentally the workings of the weblate. But still, this is something that should be feasible and not too much. But it's not any more important kind of stuff. So this, I think, is my impression of the most apparent drawbacks. But of course, there are also a list of minor things that are annoying. At least those have been reported by those people who are testing the weblet instance yet. For example, one thing is why you can use alternative languages in addition or in replacement of the English source string as a source language for your translation work. There is no copy button like there is for the English source string. But similarly, Google offers the way to copy some updates from the source at the position you are in, and we have reports, and the weblet doesn't have it this way. And also, the weblet's own idea of languages doesn't matter, it isn't perfect either at least for those more or less, I don't want to call them minority languages, like the lesser used languages and the least languages that are not coming to be encountered in the year-to-year project. But this is my thing, I think. And one of the, yeah, not, I think it's more important to have the UI working on all kinds of resolutions, it was a request to have it less waste for it to see more information on the web interface, it's trying to save the screen. Infotranslating itself, there's the send mode, it basically hides lots of UI elements, but unfortunately we cannot use this mode to accept or make suggestions. And also one minor thing is that you cannot manage your team yourself. So if you are a team leader, whatever the leader means for you in this context, you cannot grant them commissions to do translations. But since we are already having a full set of language teams, this can be easily made available. So at least my way of dealing with those annoyances is like this. So for being able to copy the alternative languages, it should be pretty easy to add an additional JavaScript and train it to be a template template to just copy the functionality from the image source train to the alternative languages. But so I don't see any problem with that. Same for the copying of internet apps, I think it is also going to be handled by some custom JavaScript to prevent this future. Same for the translating of languages, we just fix the language database and then the UI to wait for the issue is a little harder. Of course, UI design is not my forte, but we can surely change the element of this. We have the key ID, this is the project and it appears on all screens. And we can surely make this a less prominent place. There's no need for it to take up the whole line and be interface like that. So we can condense it a little bit. And of course we want also some kind of TDF branding, a leap of branding on the side anyway. So people are getting a CSS, a little bit better for our needs and make it easier to contract the database to work on smaller screens. Unfortunately, I don't have a solution for the send mode yet, but it's not just us who are noticing that there's no such action in terms of send mode that's already activated, but there's a tracker for this. But I think this feature will have to wait for the database until the web late upstream actually happens. At least I don't want to touch this one myself. And regarding the TD, I cannot manage the groups themselves. As I mentioned, we have already a complete set of languages or those who estimate coming via the mailing list to a new member or to place a banned project by a new member can all be handled by a web late who decided to illustrate this quite easily. So I don't think this is something that is urgently needed for our project, but as you can see in the send mode issue, there's a question on the issue tracker already to have this implemented because it's a useful thing to assign privileges by a code language and not just a project. And then to this can be easily handled by group systems you can set up, but of course then you still need to have a way to manage the teams Otherwise, at least when you're starting out, the new set up will be a pain to have all the requests on any of the containers, which are the assets that it shouldn't be a problem for us. We are lucky enough that we are already alone for quite a long time and we have a community already. So this is basically my view of things as I got them by reading the feedback on the mailing list and this is the point where I say no, I missed this or that and what you said is wrong, we cannot work or it should be done differently. If you have any comments, any feedback? Correct me if I'm wrong because maybe I didn't say so. I was translating some of the sentence on the help and I found that even if I have not exactly the same translation in the memory, it doesn't give me a suggestion of a slightly different sentence. That means Puto was able to, by similarity, suggest a translation memory entry. I didn't see that on Reblink. Reblink also has a translation memory system, but it is a slightly different one. So it might not be the same amount of flexibility in terms of what is missing and what is still ahead and also puto is the advantage of having the previous projects already in the database while Reblink can start out with just the master files and the existing translation, so it doesn't have a large purpose of similar translations to puto data for. Do you think that once we evolve doing more and more translation, this is actually good? It will get better and we can also add the whole translation memory that is going to be, it's for that and it's for that to puto, so this is also something that will be perfect. It has a dedicated machine translation section just like Puto has where we can incorporate other services, like for example the Amagama translation memory website, or maybe even the one that was suggested now is an extension from eProfessor. You can probably run this system a little bit, so that's work. And the translations from there also offers integration to the Microsoft terminology, but so far we are not using it, so this was also fostered equally, but then you can add additional resources, machine translations or translation memory can be accessed and created. Also, Amazon translation services, free translation services, we are not using any of those kind. Suggesting, we were talking about the people working offline. Wouldn't it be an idea that we get for a string when it is a string or model when it is taken offline by somebody, that we get an indication that it has been downloaded by that person and when it is uploaded again that it is disappearing? I don't think this is possible because downloading this available for everyone not just because that it is this and this will be hard to track and also the purpose of why someone wants to buy this is not needed, but you can add some instruments to it and maybe indicate that you are working on it. I think it's better to have it on an okay mailing list or whatever. Yeah, yeah. You can use it for whatever efforts to split up your work from this. To complement this question, if two persons are going to update the same content, will it be detected at some point in time? If we try to upload the same file then very likely or if I update template and someone translates the same file then it will be resulting in a much conflict and then I have to but can two different people or persons use, translate the same file? In the web you are in no program at all and if you are using the web you are and then you can translate the same thing. You can have two people translate it. If you are translating offline and you translate different strings in the file then it also changes the same string. The second one to upload will get changed. You can carry on talking for a few minutes while we set it to the next. You can show the interface or Kegel just tell us a bit why it's Zen mode and so this is what be the regular interface. You have alternative languages so I configured French and Japanese to be listed in relation to English. This is the key ID identification that was listed in this comment and the English source string and the intervention. This is a particular item of the special cast, which characters you can set up in your preference and basically with the fuzzy flag indicator and the Jeff button or I want to actually translate it. And here is the machine translation section where the similar translation will appear or other differences of the string in another project. In the Zen mode can be toggled to the upper right or can also be set as default in your preferences and then basically you get a less distraction mode thing and also a small focused on key control buttons. So if you use a translation press control enter just enter the next string. So as we have a mouse remode you adjust the user specifications and probably you wouldn't use a multiple source between themselves very much in this case. A bit over time now so. Thanks a lot.