Open Feature Project, 5 Jan 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: OpenFeature - Project Meeting, January 5th, 2023

Description

Meeting notes: hhttps://docs.google.com/document/d/1p...
OpenFeature website: https://open-feature.github.io/

A

B

Thomas, hey David.

A

C

D

Headphones such a disconnect in the last two minutes.

B

Sounds okay at the moment all.

C

B

Hello: everyone, hey James, hey.

B

Hey Andrew hi Ben.

C

All right just uh joining in and a little bit interested in uh trying to contribute a little bit to open feature this year. Oh.

B

Perfect welcome.

E

B

Hey Kevin good morning.

F

Hi good morning.

B

All right so Todd did volunteer to be the Scribe, but he is on vacation today. So I don't know if we have another uh person that wants to do it, um but I'm sure Todd would appreciate it.

E

I'm I'm happy to see no problem. Perfect.

B

All right I will send the link out again in the chat, hi Laura.

G

Hey how's, it going good.

B

How you doing pretty.

G

B

All right, hey, Skye, I'm, just gonna, throw the uh the dock in the chat. Real quick, if you don't mind just adding yourself as a participant and then Ben just volunteered to be the Scribe. So thanks for that, then just give it a minute or two and go ahead and get started. If you have any agenda items feel free to add it. um If you do have to get going, you know before the end of the meeting feel free to just add it to the top of the list. So you can make sure to that.

B

We get to your agenda items.

C

B

Right, let's go like one more minute: let me go ahead and get started.

D

Yeah I thought we should uh I also reposted the the link to the meeting on LinkedIn and we I saw. We should have like other three people still showing up. Okay, I think also someone from Spotify is gonna. Try to make it.

B

Okay, cool I know that there's a meeting schedule with them later, I think Pete and Todd have that on the books for either later this week or next week.

B

Cool all right, I think we can go ahead and get started uh like I mentioned. I did put the the um Doc in the chat. So if you want to go ahead and add yourself as the participant and then if you have any agenda items, feel free, otherwise I'll just go ahead and get started um first agenda item: if there's any volunteers for scribe for for the next meeting, so that's in two weeks: um it's not you're interested. Let me know if not, uh we can find someone next next time.

B

All right, we'll uh we'll Circle back to that in two weeks, um all right, so the first uh agenda item um I just want to give a quick update on the current state of client-side feature Flags. This is something we talked pretty heavily about end of last year.

B

um There's been a decent amount of activity since then, um but the holidays definitely impacted uh productivity. A little bit but I'll just provide a quick update. uh First one is I, did create a high level roadmap issue for that um there's a link in the docs here. So if you want to take a look at at basically what that all encompasses, um but basically we've done some research tasks looked at. You know how the various vendors are. You know accomplishing client-side feature flagging.

B

A lot of them have little subtle differences, so those are all documented as well, and then Pete and Todd have done uh basically been conducting interviews with different vendors and companies that are using feature flags and just trying to get an understanding of of like what the requirements are. A lot of that's been documented in the the you know, sub items here, so the the spec related changes in research there's also been a number of open feature, enhancement proposals which is kind of the first stage of spec changes.

B

So you can see I've listed four in the docs here, those ones you know the sooner people could review it, the better just to make sure that we're on the same page, the easiest ones would probably be like the provider disposal, functionality, I think those ones are pretty non-controversial and something that we should get in there.

B

You know relatively quick the one that Todd added recently is that single context Paradigm, and that's really, where we're starting kind of to lay the foundation and make the distinction between like what we have in server side, which is kind of like multi-context awareness versus client-side or mobile, which is we're kind of looking at as like a single context, which can behave a little bit differently. So please take a look, especially that ofep, because that will, like I said, lay the foundation for spec changes which, uh once you know once these o-feps are merged.

B

We're going to go ahead and update the spec, so yeah feedback on that would be very much appreciated, and time is kind of critical here, because we're trying to move quickly uh when it comes to updating the specs. So we can go ahead and get client-side support out there uh yeah any questions or feedback in any of those items.

B

Cool I I think that it's a pretty exciting uh area, so I think client-side feature flagging, Mall, uh yeah beer, obviously a huge uh benefit for people and and certainly easy to demo stuff. So you can show the value of future Flags a lot easier client side than you can server side. You know in a two second demo, so definitely looking forward to getting that out there and it's looking very good so far appreciate everyone's help.

B

uh You know thus far uh next one it's very related, um but it's basically I did go ahead and update the project roadmap board. So I basically removed a lot of like the kind of low level uh issues that we had there kind of the stale issues and just try to create higher level like abstraction issues. Essentially, so, if you look at the project board, um it has been up updated a bit.

B

The ones that are still on the backlog have not been I, guess groomed, yet so they're, basically just like kind of placeholder issues, um and so uh yeah take a look at that. If you have any feedback, certainly let me know um the one thing that I I hope is not controversial, but I ended up just rolling with it. It's like we needed a spot to put kind of org-wide issues, so I chose the dot GitHub repo, which kind of has a special meaning in in uh in GitHub.

B

So, basically, a lot of those like org related issues um have been just thrown in there uh and that way we can basically create like sub issues and Link them in across the different repos. So I'm yeah, if you have any you, know, concerns or if there's a better spot for them, I'm happy to move it, but that's where I decided to put it for now. Just because, uh if you don't have an issue created, it actually shows it's like pending on that project board in GitHub. So it's a little bit strange.

B

um So that's why that was done uh and then I did want to just call out there's a few areas that that we have issues for right now that if anyone has you know interest or some cycles that they want to, you know help out with the project. um The first one would be otel hooks. We have uh about half of the languages uh already have have a hook defined, but there's a few others that are up for grabs.

B

So if you want to look at that issue, you'll see the ones that have you know no assignee yet and if it's something that you're interested in helping out with uh you know, certainly uh let us know and we'll sign it to you and you can go ahead and help us out uh it.

B

The hotel one should be pretty straightforward and there are some examples in different languages, so hopefully that one's not a huge deal, the other one that would be super helpful, is getting started docs right now we have one for Java and and JavaScript, basically specifically node, but we'll need one for all the other languages and we can just follow that same template.

B

So, if there's anyone that has any interest, especially in a specific language or something you know, certainly let me know there's already issues there, so we just need to assign it and go um next. One is python, uh it's it's getting closer, but if there's any uh anyone that want to basically help volunteer to kind of take ownership to get it across the Finish Line this one's up for grabs as well. We are making progress there like I mentioned, but it's something that like.

B

If we had one you know kind of core maintainer, that's able to kind of get it across the finish line. That would be super helpful. uh I just listed the vendor provider. So, if like anyone's using a vendor or is a vendor and wants to, you know, contribute providers, uh you know feel free to. Let us know, there's a project board that that links to so there's. You know, like I, think 80 some providers or something listed on there, so lots lots of opportunities to build. You know different Integrations and then probably the last one.

B

This will be a pretty big task and probably take us a little bit uh early this year, but kind of just a website, refactor kind of making it uh look a little nicer and a little bit easier to navigate and try to you know, sell the the value in a more straightforward way on the main landing page and and probably consolidate the docs, and the main page together would be part of that as well, so feel free to read through that issue and comment.

B

If you have any questions, uh if that's an area that you're interested in you know, certainly let us know could use the help any questions on that all right um next one. uh This is just kind of a heads up. I thought it was a kind of an interesting use case um we wanted to in the open feature playground we wanted to have that uh like Downstream micro service um have basically feature flag awareness as well, um and so that requires evaluation context, and we didn't want to have to update that.

B

You know request to the downstream service to include things like you know the targeting key and very feature-like specific information.

B

um So, for this kind of proof of concept, we use the otel propagator, which basically automatically adds it as a header um it. ha It happens automatically as part of like an HTTP request and actually rips it back out and, like you know, uh puts it back on the request on the downstream microservice. So it's kind of an interesting way to potentially pass evaluation context across microservices that wouldn't normally have that type of information, so I definitely treat it as a kind of a proof of concept, but I think it's a pretty interesting.

B

uh You know you know technical tool, essentially that we could have for these like highly. You know, distributed like microservice architectures that that would like to take advantage of you know more advanced feature. Flagging uh functionality uh and the last one for me um is I, did start experimenting with like managing feature Flags. You know via git Ops, that one, uh if there's anyone, that's interested in that that topic I know I reached out to you. Thomas and I owe you a response on slack.

B

If there's anyone that's interested in in basically, you know kind of working through this problem. You know, let me know, uh I think you know a lot of the current get Ops approaches are, are interesting, but don't necessarily scale or work great with feature Flags just because of the number of feature flags and how they vary across environments, and things like that. You know we tried to do it in yaml and it's kind of a nightmare.

B

I've experimented with kind of like a plumy style like code Generation, Um so I'm kind of working through a couple different ways of actually doing get UPS feature Flags, but that's where I'd kind of like some feedback and to bounce some ideas off some people. So if you're interested um you know reach out on slack or you know, add a comment in these meeting notes, and maybe we could, you know, set up a quick little uh sync on the topic and kind of discuss it in more details and that's basically it for me.

B

It looks like David's next and before I hand it up to him if you get. If anyone has anything to add to the meeting notes, certainly uh you know feel free to add something. uh I think we're a little light on on topics today, probably partially because of the holidays. So that's it for me. Go ahead! David all.

D

Right thanks yeah on my side, it's more of an FYI. We just started with the incubation proposal. It's still work in progress uh feel free to have a look, um I think Mike. You had started working on the technical side uh aspects, sorry, but whoever whoever's you know willing to help with you know any any contribution will, as always be welcome. uh I just filled in some of the community related items.

D

So if you have any questions there or any suggestions or think anything's missing, you can feel free to um yeah feel free to leave a comment. uh That's pretty much it!

D

uh The idea is hopefully, if we manage to get this before kubecon and before they actually um open the the proposals to get community days, we could try to have a community day at kubecon EU. Otherwise it will be for kubecon, North America.

D

And that's that's it.

B

All right cool it looks like uh Thomas added one. So if you wanna yeah.

A

I just wanted to ask if you want to have like some uh toolingo for their experiences in uh in this project, such as like I, don't know tools that look at other repository and say: hey this flag is configured in your code, but not everybody in a provider or something like that and I.

A

Don't know if it's something that we already have been about in for open future or is it something that where we don't want to go yet and we'll learn today as this kind of stuff, where you can run this CI script and candidate if your flag is in your code, but not in your in your solution, and since we have like a more standard way of doing it, we may have something like that. That is instantly interesting. Also in a project.

B

Yeah I mean I, think it's worth, maybe exploring it, maybe a little too soon to dedicate it a ton of resources to it, but I think it's worth considering for sure. I did notice. I think it was a Dev cycle had a pretty cool, uh GitHub action that basically would show you, like you know, based on your code and based on your configuration in the tool, it would show like kind of the diff. What's been added, what's what's not available things like that?

B

That would be pretty interesting to kind of standardize that, if possible, but then you'd also have to figure out how to like interact with the source of truth, whatever that may be, which may be the challenge, but um you know for us, we could pretty easily, like probably detect, which feature flags are defined in code, but then how to compare that uh to where that's uh you know where that actually, like resides, may, may prove challenging, um but yeah I mean I.

B

Think it's it's worth exploring I, don't think there could be a lot of possibilities there and it certainly makes the pull request more interesting. If you could have those kind of like auto-generated comments about like how it's, how this flag's being used and what the impact would be of, adding it or removing it.

E

E

There's probably a class of features that you could achieve just getting some extra metadata out of the provider right like you could achieve that, potentially just by getting a uh list of flags available within a project or something from a provider uh like an exhaustive list. That might not be what you get back from um the default provider, um but then there's queries and questions that are coming up in my head around permissioning um and things like that, because then you can go the step further and I mean we've.

E

Had a we've built, a proof of concept where um you can attach an issue apply to a GitHub issue, and then it creates a comment that you can toggle the flag.

E

um You could potentially drive that through her open feature, but you start to get questions around.

E

um Yeah like like sensitive Keys, um because they're doing potentially volatile rights and things like that, yeah I.

B

Mean we haven't.

E

We haven't talked about that at all right, like that. That's kind of interesting like it's, it's probably just simple metadata that you can expose I, think.

B

There is like a a proposal that actually Beyond The Proposal stage. Now there was a an offep that Pete had opened about uh provider metadata um so that that is something that's uh there's a pull request open for the stack right now, it's not been merged in, but it is open.

B

um I think the way I interpreted uh Thomas's point, though, is like you basically have a tool that would show you like a list of all of the flags that was, you know, available in the code, and you could probably you know, do a comparison between. You know uh some commit hashes and things like that and see which ones were added and removed, and then you could basically use that information to look at like the source of Truth, like you know, flagsmith, for example, to say like do these flags even exist?

B

um Do they return the correct? You know type things like that. You could basically like give people the confidence that that this is going to work. You know essentially before merging in the change.

A

Yeah yeah, that's exactly my point: it's like because uh and you're probably right. It's maybe too soon to think about all of this, but I think it's a good next step to improve usage of your flag and being sure that people are tooled up to manage your flag in a long time in the long run. So that's why I was asking for that. It is because we did an entropy.

B

Yeah I mean one approach could be just to have like if we had like a kind of a way to detect the flags in a generic like open feature way, and then you'd basically have to kind of do another like provider style thing. That would be like the vendor that you're using to to verify what the source of Truth could be one way to to handle that um yeah it's worth exploring, but that's something that you're interested in I. Think.

B

Definitely, uh you know, take a look at it and let us know how we can help, but.

E

I do I do remember. A while ago, looking at the uh I then looked at the source code for the launch, dark tool and I, because I I was thinking it might be doing some crazy.

E

um uh uh um You know looking at the language and trying to evaluate and build a graph of it just it. It was at the time anyway. It was just searching for string, literals and flag names that matched um so you could Pro. You could probably get a an interesting proof of concept with a relatively small amount of code, but I think what would follow that what would fall out of that would be quite interesting because it would just be.

E

You know like these: are the metadata entities that we need to be able to drive this to to to solve this feature and that'd probably be fairly straightforward.

B

Yeah I mean I, think they're like there's two other parts like if you could tie it potentially into the get ups and then that, like compiler, aware of feature flag stuff that I kind of showed off a few weeks ago, like all those things, could work really nicely together potentially, but it really it depends on the org too I mean some people wouldn't want to.

B

You know have to deal with that that level of everything as code, but the teams that do want that could really I think benefit a lot from from that whole network flow.

B

So something to think about. Maybe that's how we could tie it in Thomas. If we we talk a little bit more about like the get Ops flows and stuff and like the compiler awareness and how that could kind of work end to end I think would be a really interesting uh flow and and very valuable for the teams that want to embrace, like you know, kind of type safety and those that confidence across the whole life cycle.

B

B

All right, um yeah, like I, mentioned a fairly late meeting today, so if anyone else has anything to add feel free to to chime in uh otherwise we can probably go ahead and wrap up pretty soon um like I mentioned. If any of these items are of particular interest to you, you know, certainly let us know um you know, because we could definitely use the help for for all this stuff.

B

B

I guess we did kind of skip the part uh the the new people introduction. So if, if anyone and it feels comfortable and wants to just introduce themselves and say hi, uh please feel free to do so.

G

Hey I'm Laura I'm visiting from debcycle and I actually built our GitHub actions, so I can help with any bringing any of that to the table.

B

Oh cool yeah nice to meet you.

F

Yeah, maybe from my end, so I'm kind of so working for dantrays and yeah I'm, trying to see if I can contribute and yeah try the topics so yeah.

C

Hi I'm Andrew I'm from gitlab I work on the future flags at kit lab so interested in in helping out as best I could here.

A

Perfect thanks.

B

All right well, thanks for uh introducing yourselves, appreciate it nice to meet everyone and uh yeah I. Guess if there's no other topics, we can go ahead and wrap up, and if you have anything throughout the the next few weeks, uh you know, let us know in slack and if you haven't certainly feel free to join, um pretty pretty uh lively bunch. I would say, especially comparatively to uh the rest of the cncf uh slack channels. I think this one uh has a decent amount of activity.

B

Cool all right thanks, everyone hope you had a nice holiday, happy, New, Year and.

G

C

B

All right goodbye.