 Should we? Let's go So, yeah c++ 17 I added this one I think has someone asked them slack about it and it's kind of like one of these are perennial questions When can we get c++ 17 and you know the motivation there comes from? Obviously like language-level features, but also just like they've been proof standard library things like file system being standardized and so on and Last time we discussed this I think it was just the compiler support wasn't generally there Like anything across the board and clang and GCC we felt that they had been sufficiently well baked versions out there for long enough Have things changed? Probably yes, but now there's a there's a new concern which is You know, not just for the mobile project which are doing which requires Android and iOS support But people are also I Think people are starting to compile envoy on places that may have less modern compilers. So I Think what we need to do here And this was an action item that I had from a couple months ago But I never actually did so I'll assign myself an action item is I think we need to do a Survey and actually try to better understand like what OS is people Are using what compilers and and whether you know We can determine what what the tool chain support would require And then I think we have to try to figure out where people are actually using on boy So could you actually assign me this and I will get a survey written that we can collaborate on and then we can send out And and I can also have someone look at it here here at Lyft on I just I I Dispect that Android and iOS are both behind on support So like it wouldn't surprise me if they don't fully support C++ 17 yet. I mean I always do not Sorry, I Was just gonna say they do not we have a couple Android projects that Cannot make the jump with the rest of our code base to C++ 17. Yeah, I was the one who proposed this flag So Yeah, they they do not Yeah, our our experience so far and starting the port has been that there are some surprising things that are that are missing so Yeah, that's my guess too is that we will not be able to move or We be able to move but not but we'd have to require that we don't use certain features So for example, like we can't use the file system API or something like that But if it's okay with people I think like I will take the action item to start this survey because I just feel like our user base is too wide at this point to really understand what the implication would be and It's the kind of thing where there's lots of people that use it But don't come to these calls or even track get up and then we'll make the change And then there'll be a bunch of people hopping on and saying we can't compile it anymore Okay, and sorry and I thought for a while that even at Google you didn't have C++ 17. Is that not the case anymore? I Think we're heading there pretty quickly if we're not there already Yeah, but I mean this would be a blocker for you unless you act like you explicitly I mean, I don't think we're trying this issue talking about turning this on like tonight today, you know This is like Will we in a quarter as to when will we be there? Will we in a quarter's time be at C++ 17? Is this thing if it's a year off this kind of thing? It's essentially what we're interested in learning in Okay, all right, but I mean it's like you I mean You as in Google need to fill out the survey also because if you don't have C++ 17 for six months, then you know I mean that's that's a that's a data point. I mean an interesting need to ask in the survey will be You know, what are you using today? What you anticipate using in six months? Yeah, like that like yeah, the temporal aspects of this. Okay. Yeah I will I will work on this Survey because it's something that I think we need to do anyway, and I know it's on my list Which is ever-increasing and never decreases in size All right, I just moved to the top of the list. So maybe it will it will go faster Okay, did anyone else have any comments on that C++ 17 Okay, UDP who was behind the UDP based DNS filter item? Oh, yeah, that was me Okay, can you speak close to the microphone? Oh, yeah Can you hear me? Yeah. Yeah, okay So I think this topic had come up on github a few days back and We added the UDP support recently and I saw quite a few issues Back to my pull request and to the issue I had opened initially So I wanted to discuss on the call if if there is enough need for us to actually go ahead and add the UDP filter or the DNS filter into the envoy Extensions itself and if we have to work on a way that everyone agrees on how it should be done I I don't have a strong opinion. I've heard like I mentioned in the github I I've heard of some people wanting this behavior So like it wouldn't surprise me if people would use it But I haven't I haven't explicitly heard that someone would would be working on it soon I know that the SDO people had mentioned maybe wanting this to respond to DNS requests for service discovery um so So I have so even we wanted and so that's the reason so I'm already I have started working on it on a private filter repo Uh, do you think it would make sense that I go ahead and do that? And then we look at how we can bring it back into the code base uh after After the fact that it's developed or I think that's I think that's fine. I guess my opinion here is that This seems generally useful So it's like it wouldn't surprise me that once it shows up in open source people will find interesting ways to use it at the same time because I haven't heard an explicit request for it You know, I I don't want you to like waste your time just doing general open source work for no particular reason So I mean one option maybe just to do it in a different repo But make that repo public and then you can link back to it so people can look at it And then, you know, if there's interest we can just get it upstream. Yeah, that makes sense So that's what I figured and I started along that approach. So I think my repo is public right now. So Yeah, so maybe just link to your repo in the github issue so people can follow that thread Uh, and then if there's interest we can we can get it upstreamed Okay, makes sense. Yeah, or or if you know it You can make your own decision later in a sense that where The benefit of getting it upstreamed, of course, is that then, you know, we do all the refactors together and we fix the bugs and stuff like that So, I mean, you know, there might be There might be benefit for you to getting it upstreamed, but I'll I'll leave that to you in terms of what you want to do Okay And is there any Is there any examples of how extensions? Use the external dependencies that envoy itself has Yeah, so there's the envoy filter example repo, which does basically exactly what you what you want to do Okay, no, the only for example doesn't actually use envoys External dependencies He uses envoy as an external dependency and so right But I mean, I think like Many of envoys baked in filters are making use of external dependencies which live in the Proper like the thrift one or the Various grpc ones. I mean, there's not even the same repo, right? Yeah, but there's not that much difference when you're in an external repo really Basically, like as all flattens out the namespaces Basically a workspace So basically all envoys external dependencies appear As top-level external dependencies of any project that consumes on boy Okay, okay, that sounds good because I'm planning to use the crs dependency. That's already there in envoy and I wanted to see if I have to add that as an external dependency to my repo separately Or can I just use it since it's already in envoy? You can just use that Okay. All right. That sounds good. Yeah, once it's once it's loaded in you should be able to refer to it using normal normal basal notation Okay, okay Yeah, I think that's it from my side then Okay, do you do you know lizon whether the sdo people would actually So I think I think that idea came self a couple times, but we don't have that in concrete roadmap or like Any like timeline yet. Okay. Yeah Yeah, I mean it's the kind of thing though where I I I just suspect that if the code was written and working available people people would people would use it That's right. Right. That that's that matches my guess as well. Okay Yeah, so so maybe just get the code working and then drop a link in the repo and then if there's interest or if you want to We can get it upstream Yeah, it's um, it's uh, michael here from jp morgan. I know the pivotal folks were looking to get udp support Uh shannon cohen If you go back through some of the issues might be interesting to check them in the pivotal folks I know that I we'd expressed I'd expressed interest too But our sort of use case for the public cloud has sort of gone away now But uh, I'm sure that we would be interested down the track as well Yeah, I mean I I I think the other use case here that your pr unlocked is that It's actually will be I think relatively little work to do upstream udp now So, you know, if someone is out there who wants to finish that work we can now do and and to end udp proxy I think it would be not very difficult actually Right Cool So do everyone just enter their own names in the attendees list or is someone updating that? Oh, you can just type in there if you want. I mean it's It's optional, you know, I see, okay, you don't if you don't like to be tracked Cool, do we have any other agenda items from the community? I don't I can't think of anything. Uh, the only thing I Will mention is that we're getting ready for planning for envoy con 2019 So hopefully we'll have the CFP open by the end of next week Maybe and we'll be looking for people to join the program committee So if you're interested in reviewing CFPs, feel free to reach out. I will add that to the agenda Is there a Registered like limit on how many people can attend the event Yeah, there's a limit. So, um last year we sold out And pretty pretty fast. So I think we underestimated the demand I think this year we're going for slightly bigger venues So we'll be able to probably have about 500 people. Um, so hopefully it It won't sell out quite so fast. It's going to be co-located with kubcon and san diego So it'll be the monday before kubcon. I think it'll be like november 18th Or something along those lines in san diego Um, cool. Um, I'm trying to think of other general announcements. I can't think of anything Did anyone out there have any questions or Things to chat about? Okay. All right Call it early. Bye Bye