.Net Foundation .NET CoreFX Contributor Standup, 22 Sep 2017

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: .NET CoreFX Contributor Standup

Description

Link to discussion: https://github.com/dotnet/corefx-standup/issues/29

A

B

Know that I, muted myself, it should work.

C

What did you call it Cara when I go over this to type? Something just goes by that I.

D

Think I can type it so. Okay, something okay.

C

You guys here as an own firm that.

D

We can, you can hear us, so that's great.

A

C

Anyone got audio now, can you guys tell us if you have already pray all right.

D

Beautiful sorry, for the rocky start, that happens all.

B

Right, let's start yes, so let me probably the.

C

Best in Rio introduce ourselves I think we.

D

Can be that idea? Actually so before we introduce ourselves? Actually, you know we'll come to the first contributor stand up which is live. We have made it in 15 minutes next time, I'll be just it's gonna be 13. Next time, I'm gonna go down, and before we start introducing our stuff, I would just remind people that the goal of this end up is really targeted at contributors of core effects repo. We want to kind of open the bi-directional channel here out our contributors. What are their ideas?

D

What are their pain points things that we are missing and you know how we can help them to be successful in our repo, mainly core FX. At this moment, logistic realize we're gonna, be monitoring the live chat. So if you have any questions or something you know put it there and whatever will be easily, but we might answer on the spot and rest you can all leave for the last part of the session. I think it's time to introduce ourselves. So then.

C

I am Dan I'm a dev lead on the base class. Libraries I'm, not the only one, but so I I own, a bunch of things like the collections and primitive types, all right.

D

No need to introduce mo I, guess or maybe.

B

Probably, what most people know is Don s tanner, but I was working the basis of libraries. We expect features like yeah.

A

I'm Wes Wes a great former deadly. Now engineering doesn't let standard done in libraries, which everywhere around on that course X in particular,.

D

Alright and I'm Carol Sigmund I'm as well, delete on the BCL class libraries or on bcl and I, try to step up and draw a bit of community management, so you know make sure that our contributors are happy.

D

Alright, okay, so that's it for introductions and the next step you would like to you know: we prepared a kind of brief list of things that are happening conqueror of X repo for people to know again, if you have quick questions, feel free to fire it. If the answer might take longer, we might leave it to the last part just shown up at the link to this in the chat. Maybe the questions are not there. Oh yeah, you know, and it's actually in the feed or Jordan, so we prepared a you know just to say.

D

Like brief, you think things that are happening right now in the repo. So, first of all, you might have noticed that weirdo, you know kind of focused on delivering great to.net core 2.0. So we shipped that, probably more than a month ago, one and a half months, I believe and we're still looking at things that we broke by accident, I'm, you're, trying to ship them in service increase or service increases.

D

You should expect that there will be rough play, or you know, two months period of shipping servicing updates and it's kind of called for action for all the contributors. If you guys notice that something is broken in dotnet quarter, ATO against either previous versions or against desktop it's something that we definitely want to know about and consider for fixing and I. Think guys. You won't read you ever think the team is actually or non-trivial part of the team is focused on built infrastructure, kind of the build from source effort, which is not quite ideal.

D

Maybe best can say couple of things about that.

A

There's a attention we have like a source, build repo, that's occurring and I'm going to be uses our base sort of thing. The other fold on that Cortana and going to some I'll, do all the other repos that make up the food that make more products.

D

He's like 20 or 30, or something like that. No trivial, number, right or ones and.

A

Under the sdk ones, the core month, basic or set up work backs cor cor standard repo until I decide that the goal is basically be able to build full COI SDK piece from source. So we can easily bring up the distros and other such things and insert into district district. Actually so.

D

You will see changes in that area. It might impact actually even the core effects infrastructure, how we built and thanks and this project will probably run over at least a few months. Like you know, it's good, it's okay,.

A

D

Alright, the other thing that is happening or happened recently I think it was this Tuesday. That's at the dotnet called V announced windows, compatibility fact for dotnet core, which is actually you know, heavily impacted on the core FX repo on this slide from Scott hunter. We had things like that: we're gonna ship, a pack with Microsoft inserted registry called Dom configuration drawing and other things, I think that you know eventual among them will be directory services anyway. So this is coming. The preview is already available right.

D

C

Yeah went down Tuesday perfect. The preview contains two parts. One is everything that we already shipped in 2.0. There wasn't in the shared framework, so you don't get a free reference to so it's already available and you get, but it you had to find it so that just pulls everything in for free and then the other thing it has a system drawing which, incidentally, also includes the an evolution of the mono system, drawing implementation, so it works. Are you next to and then for final release?

C

It'll be later this year will include a whole bunch of other things. We know some things will certainly be in there like perf counters event, log system, ODBC and a bunch of others, and then some we're trying to make happen. We don't know yet about system man in German system directory services. We hope to have in there and then there's a few others so, and that should be later this year.

D

So yeah, if you guys, want to jump in and hope that some of those things if you're interested in them this is the right time and actually on some of these things we had actually some you know significant contributions from community, especially on drawing, if I, don't, if I remember correctly, I'm, not sure about configuration and actually some of the on the directory services as well all right. So three more things that are happening on the repo I have here. So you might have noticed that we have some.

D

We, you know kind of created, a backlog of API reviews over the summer. We burned that you know fast. People are on vacations and things, so we are now like at 12:15 or something like that. So we will be catching up over the next couple of weeks or a few weeks, so just FYI that it might take a bit. If you see something that is really you think is critical. Let us know overall, I believe that all those API should be you know approved in the next couple of months.

D

So if someone votes to get them into two point, one that I didn't see a problem of that again I come from and that will be actually approves. I can promise that they will be review. That's actually different thing.

A

The other address of the questions in the chat sure one of the questions was, it is I. Think Johnson was asking whether or not our source book stuff is going to help, but the PS code having contributors I'm using Leah's code, yes yeah so I, think.

E

Yes, we do have BS.

A

In thousands, for obviously predominant court, like absence of corvex, is a little bit of a different beast, because we don't use standard projects. We are hoping to move to more of sta breeze projects as we get and we can actually start consuming the two of our SDK within corvex and I. Think that will help a lot. What's the tooling for the projects for other platforms,.

B

Another angle, actually on record explaining why that is like the one of the problems that we have is that, because we are building the platform right, you can't quite consume the park and, like our customers do so most of the tooling that we have makes assumptions around how the system is laid out and like who's shipping ourselves with that public schooling is a lot harder. I mean highly the repos, like asp, net or rosen, basically compiled the same way that the customers compiled have a much easier time getting to link story off the ground.

B

But I think it's fair to say that it is always been a pain point or team that effectively you can use ultra edited sublime text. It's not really video.

D

But to answer actually the question about: like you know it, you know, I think the key root question here is, you know, build vs could be able to be used for developing core effects so Dan, actually on his on his whiteboard. We have lists of like 20 things that we want to invest, and this is actually one of them.

D

We discuss it just yesterday, so we hope that in we will be able to put in next month or two someone on that to kind of dig little bit deeper and make it more at least better for better core contributors to use vs code.

B

D

Particle, though,.

B

D

Know, actually it works and I mean I, think the John's I know they told us that it doesn't.

A

Work right tons of targets, another thing using loading up our stuff and vs codes to process.

B

A

Barely works in vs was time.

C

To so Jonathan I assume you're on Mac right, not Linux, I'd be interested in any other feedback you have offline about. You know the development experiencing core effects on Mac, because some people do it here. There are some people who work primarily or exclusively on Mac, but most of us work on Windows, and so sometimes we can be a bit short-sighted about the Mac or the experience.

C

A

Well, it's just a drawing. We implement, we just released a beta.

C

Oh yeah yep, it's done I, there's, there's actually a bug which prevents it working on Mac, but that was a typo. That's been fixed, so the next release world out there working bread, I know of no significant bugs or significant work remaining on it. It is marked a preview package, but it's it's good to go it's available. So.

D

Do other cool things that are happening on Cerf, actually Paul, you have might have noticed that span of T and friends are kind of being. You know we're trying to kind of spread it around incur FX repo, adding bunch of new IP eyes in that regard, and the is to have actually you know significant part of that done in 2.1.

D

There's actually call for action for contributors that you know we could definitely use some help and implementing some of these you know specified api's, incur effects, and the last thing you might have noticed that we are lately doing actually larger api reviews and migration of some components from core FX lab in.

D

C

All right, we have any questions or anything I would lost the sound on the pitch.

B

B

Don't know why interesting.

D

Don't touch it anymore,.

C

It's like non-deterministic, okay,.

D

Let's see it works so the next step we should probably go. Try. Five should go to the first topic that we picked or I was voted most, which is the API review process and meetings. So should it be don't project that right, so it doesn't matter.

B

D

Don't have to that's. Ok like people can process it.

C

Where is the material in there.

D

So I bring in here the cheat sheet of the shoe number 10 over here. So, as most of you guys are probably aware of the API process that we have, we have it documented in get up repo. I would say that it's you know it's mostly true. There are some details missing and those are things pro 2. You should talk about a little bit and I.

D

Think thing that key thing that mmm is not quite popping out of the document is that our API review process is actually not staged, that we rely in the first stage on the area, owners or experts to kind of think about the API. Look at it and do kind of first level approval like yeah. This looks like a good idea. It fits space might be good thing once we have that, you know.

E

If they weren't in the first.

D

Stage, actually you don't get the bigger API, you know API design, group involved and that's by design, because that group would be super busy if it would comment on very early. You know ideas for every single API and once the ones the API is kind of approved, or you know, or a proof like you know, considered a reasonable thing to bring forward the area expert actually mark it as IP are ready for review and then, when gets actually to the formal API group, and actually you can see over here in this room.

D

There are like all four of us are actually part of that group. On top of that, it's Steven Dow, first off Selena, Jeremy, Barton and probably a bunch of people, a bunch of other people, I missed who are joining us here on their.

B

Very dorans comes to mind when we discuss security at all.

C

Yeah, we usually pull in, like the experts in Scripture, we're not gonna, do it up there and so forth. Maybe it's implicit, but we could briefly mention why the API process is so honest because it is um and it and it's probably obvious but like once we have an API, it lasts forever, so we still support API to the ships in 2002. So that's one reason we have to stand by that API. We don't want to have duplication later, so we want to get it right and another is API.

C

Design is actually quite a subtle thing. You know some sometimes folks who have done this before they. You know the eighteen club I want. You know food to do bar and so just put a bar method on actually there's a lot of other things. Didn't siddur like discoverability, like composability, like performance and consistency and many other things, and I one of the attractions.

C

The of the donut platform for many people is- is that in many places we've got that right, some places we definitely didn't um and I guess we want to not lose that body problem. So hopefully we read the right balance between pain and and benefit. You know, I, wonder.

B

I wanna share like a little bit here. So what why are we? Why we bother with reviewing API is because stuff like this happens, somebody edit the API they didn't intend. This will be a public API office Lee but like they they're just headed for the time being, I think there. The comment here is actually updated a while ago, but the comment originally read obsolete.

B

This is a temporary API and holy celerity gives us an official API to detect this, and we ended up shipping. The API, because you know, but you know the framework- is large diversity, API or something, but like we do this on an ongoing basis that we essentially build sleep, the the API surface, with what we have shipped last time, just to see whether any you know any anything had sneaked up, and maybe you know you just missed it and like. If you look at the framework, as then said, I mean ever quite a moving.

B

The API is in the fabric, so it is substantially in size and think it's not that people don't think about their API. So the most part we don't get like you know terrible API proposals, but it's it's always the case that team a needs to do something and then team B realizes that all we have something there. Maybe we should collaborate and okay.

E

If you're having a.

B

Centralized review board, you make your life a lot simpler to see, essentially the set of like asks that are like consistent and then maybe there needs to be centralized API rather than two teams building the same thing, and on top of that I mean like you know, Crystal Selena wrote this book framework design guidelines which is like 250 pages long at this point, or something like that, and it has all the do's and don'ts of API design but like it is fair to say that nobody has the book in it in their in their head all the time, but it Christoph so API design is always a team effort where you poke holes in the design you, you know your question whether the API will person.

B

Hopefully your question whether the API will work properly in all cases, and we have this random joke on the team. Every say you can always ask the question: what do some things are just fundamentally complicated? If you don't see through these things, at some point like and then said, you only have one shot right and then you know the best you could do for this. Api, for example, is obviously doing it like even this API you.

C

Can never be, and we don't want to put up to the attributes on anymore. So there's that too, because.

B

Obsolete doesn't really work either and so.

C

Sometimes that discussions can get not heat it, but like pretty passionate right, but what's interesting in E is it's hard to think of a time when I've ended up leaving the room being unhappy of the results? I've usually gone, persuaded that I was wrong. Sorry, it seems to work.

D

And actually trying to be a bit more open in those things. So lately you might have noticed that we started recording some of the sessions API review sessions and as soon as the discussion is not too passionate, we are actually more than happy to release those things. So we did bond like three weeks ago and actually from this Tuesday we're gonna again release the video. So that's our way how to try to kind of give people as well idea, what's happening behind the scene.

D

We don't want to kind of open every single, a design review meeting and have it live or streaming or anything like that. The reason is that we want, like all participants, feel encouraged to speak about and I think, and sometimes it's just you know not ready to be published. Oh.

B

Don't say like it's the you know, the question is that what kind of like? Why are we doing this in the first place? I know? Why are we sharing those review meetings? I mean the the goal that we have is education, and we want to make sure that the community understands how we're thinking about the problem space. You know what kind of discussions were having, but in the end, like you know, let's not forget.

B

Each of these meetings is two hours long, usually because design meetings, you know you cannot do much design in the capital or essentially, which is basically results in like people are what come here. Look at they're coming around: it's all, oh the yeah. Basically, the problem is that you know we have.

B

You know there's still only so much time in a day, I didn't if we up upload every single video that we do then, quite frankly, nobody is watching them either, like people are not going to like watch hours and hours of us talking over the same thing over and over again, like I, think there is value and like do another one basis, but even if you miss five of them like there should be very limited surprise. It's like and I.

C

Think I think you're right, I, don't I, don't think it's a goal. We have, but there's someone to be able to see the video if they have particular API requests being discussed like that I think the overhead to us of doing that is too high. But yes, some example ones. That I think is helpful. More continue to do that. I.

B

Mean like, if you take it to the extreme, you can all buy a body cams right and like run around your like extreme 24/7, but then the question really is who's going to watch. That and like is that actually helping you, it I think in many cases, like writing up the principles and sharing the you know the viewpoints, if you have this is a much more effective way to like you know, get.

D

The point across what's key value from that I see, is sometimes when someone is interested in some API and like okay, so I wonder why they end up with this set of principles. Actually, there's a link, you can go and watch it. So you know you don't expect anyone to watch the whole sessions much more as a reference when there's interest and specific.

C

Interest you folks online have any questions just type and we'll jump in.

B

The only other thing I would add, like the other reason why I like it's community videos, is also that you can actually see a face to the discussion. That's not just yeah the anonymous guide. You know developer units 76 a commented on github issue and said the API is bad. Like I mean if its actual name and person the face I think it's is actually helping the interaction quite a bit.

B

C

Do all have Ferriss and github I think now you.

D

Do but I know how hard it is to convince engineers to put their motors down yes, I'm doing they don't wanna, see puppies. Yes,.

C

C

That's right, you.

B

All the shape no I mean, like the I, think. The other thing is that I, you know many people know chill online. Probably right, I mean sure lion was the idea that we stream a little bit like you know, from inside the Death Star to the outside world, and if people sneak peek into house off engineering, then I might consider we kind of do the same vehicle or so that's not.

C

Something that we invented right so question from Tanner who is actually on the C sharp E is.

E

On the C, sharp.

C

Team thanks Tanner, you can pop down the hallway, if you like, I, think one of the things that would help with API reviews is. It isn't immediately obvious where your API is and the list of pending reviews up. It's been marked as ready review, so it can look like it's. It's ready for review for two months without an update, sometimes I think that actually happens. Unfortunately, other bottoms say that yeah that probably you've been very polite.

C

Tanner I I think we could probably have a more formal system around that, like sometimes we sort by oldest and sometimes someone actually wants to write code that day, and so we pull that up. We, we probably could do better and I think. Maybe it's a bit of a backlog built up yeah.

B

I mean that's a German home that I discussed with Carolyn some porn I. Don't know how you guys think about that, but I look at all of our peoples with its walls in or corvax arrays. In that, like we all have at least a thousand issues and most of them it actually have more than a thousand issues and I think with like yeah. Well, unfortunately, our two thousand mile.

D

I think the reality is.

B

That I mean we originally said. We want to keep things open with the intention that you know, because we all grew up and put all days where we did this. We you recycle us and then towards the end of everybody.

B

Saturday's massive push to resolve all bugs, and then we we both resolve a ton of them and say you know what we're out of time like resolve on fix, essentially right and as an engineer, yo I hate that, because we don't want to leave work, I'm done and you wanna make customers unhappy, but the reality is also if you have a backlog of a thousand or 2,000 items, you're going to boom the backlog on an ongoing basis anymore, because you cannot just go all you know through all the pages and see the you know the vitally important things in it.

B

So it's a little bit of, like you know, getting lost in the details, that kind of thing so I'm, actually more or less of the opinion that we should be following even more aggressive with closing issues and the you know, move the down set of course that we will close issues. That may not be a bad idea. Just say we don't see ourselves doing it in the next, so that doesn't.

C

Apply to API ready for review at those guys. Ideally, we would plow through one of them. I think the last log is where this API needs work and no one's groomed that, but even even if you wanted.

B

The API is that are ready for review or like Mark for a guy edition. There was a decent amount of it's not like three pages I mean we did a lot of work to make it to three pages, but this is a similar in the sense that we basically said more aggressively. No, no or not ready and I think the problem is sometimes. The question is how much vertically one is spent, making it get ready for it. But in many cases somebody says I wish there was an API to do.

B

X I they're not actually proposing a design for X, so somebody else has to pick up the work and then the question is well. We let the issue sit here. Do we just close it and say they could come back with the design we can talk about otherwise, so.

D

Speaking of which that's actually something I wanted to mention that agreed on that three weeks ago, that you're gonna do a small tweak actually to the API labeling that you're gonna. You know right now, if you're using the API needs work for all incoming new ideas, which we might not even buy it by them yet or we don't even understand it, they're valuable and it's mixed up with things we actually think like. Oh, this is valuable.

D

It really just needs a little bit of more work, or maybe it came back from the API review kind of incorporate feedback. So our plan is textually. It's on me, hopefully the next few weeks to actually introduce a di, such as Chung label, which is going to be basically the label for all incoming stuff and part of that you're gonna groom a little bit.

D

All the API needs work right now, especially smart up for grabs and if there hasn't been honestly interest in a year or two from anyone to create a formal proposal or you know push it farther and it doesn't look like slam-dunk great idea of you're likely gonna be more aggressively. Closing jobs so expect that I.

B

Think the other motive you had internally was that you know teams would want.

A

To review the API.

B

C

Was going like.

B

It's they basically work on the feature without waiting powerful for the design. I didn't innovative, it come. You know with enough buffer to change the API, but they don't they're, not blocked by the review. I think the problem with the community is that they feel they're they're blocked by our approval, which, to a certain extent, is true, because the approval is basically what you need in order for us to make the decision that you've only ship the design for a partner team internally I mean they can shoot their own bits. They cannot tell them.

B

Nobody can only tell them that the way they have done it is incorrect, but we cannot really prevent them from chipping it, but I do think. There's a there's, some sort of a problem, I think in or way we work daily.

B

It seems like the process is too heavy way too early I think like it would be better if you could get two motor people. Can you know collaborate among themselves to the point where they feel that the design is baked enough? And it's the only question of like you know: what's the shape of the API?

B

What's the name like and is an inter phase of an abstract base class and what everyone and- and it seems like something that you can also turn around faster, then you know the discussion of like oh, should we do X or Y to begin with and I think make some of the discussion. I think CMOS a bit early, because the PD people proposing the design don't even fully baked view of what problem they're trying to solve the first place. They were Hunnish that there is a problem, but it makes a discussion somewhat.

B

Like you know, long.

C

That that's a segue to how you can make an API review it meeting more efficient, and this there's a wide range and and thanks to folks who've, been following your template or something similar helps a lot I think for us to be effective in an API review. We need the API actually written out like formatted as code, but we also need like a definition of like the problem.

C

You're trying to solve alternatives, considered umple code and usage is very important and then one thing that really helps is as discussion continues to go, update the top post. Now we we as a matter of policy except in extreme circumstances, don't want to edit anyone's post. The only exception I can think of is when we go and add code formatting, and then we put a little note saying: we've formatted it. So that means that we can generally do that.

C

So if, if you create an API issue and then there's a bunch of discussion, please don't post further down. Ok, I'm, changing the design to blasts alone go up and edit it so that the ideal API review. Meaning is you just look at the top post? It's really hard and inefficient for five or six people to like read a long thread and the.

B

Other thing that people seem to you know not be aware of necessarily, even though it may sound trivial, but it's so much a debugging right. Every time somebody submits you're at you know a bug report, it's surprising how often developers don't give the details. You need to act on the back, but the stack tray. Is there a message whatever it is? I didn't it's same of API. It's like I mean we're not do main experts in every single technology that you know our team owns or the you know.

B

The dotnet team as a whole has shipped over the last 20 years right, but like 15 years and so like in many cases we don't have the domain knowledge like to even understand your pollen.

B

So you need to do some level of work to convince us that oh here's, the problem that I can't solve today at all I have to go to C++ or I can do it today about it's like 30 lines of code, and this is what I want to do and like this is like you know how three lines of code should look like I deal in like getting that pitch right. So everybody who is not domain expert can understand your pitch in, like let's say 30 seconds to a minute.

B

I think is something that makes our lives substantially easier.

C

um This is a question to get you, but just one comment in case you don't know every area label like it's sort of area system, um time or whatever has it a couple of designated owners? If you go to the main core, FX homepage to the readme scroll down, it's like a section like issue guy to issue owners. If you click on that, it lists all the owners, so you can see who the two owners eclipse, for example, those guys or women like the people that you want to talk to.

C

If you want to refine a proposal, they're really the ones who are responsible for marking it ready for review and and so feel free to, like cite them by name so that they all turn up. I, think people pretty.

D

Usually, do I usually do but.

C

It's like, if it's done quiet, that's the way to make the issue alive again. This thing that's earth call me if, if nothing, if they don't the question from Kevin, is there was some talk about moving with proposals out of cortex to be permanent, so there and repair like c-sharp Lang, has been any more thought on that yeah I.

B

Mean I've I've spent a little bit time. Thinking about that, but nothing like up to my head, I think I mean the reason why there was a team did it was that um language, design and compiler implementation are fundamentally very different. Exercises and language is dying in itself. Is engineering like understanding the types of Steel implications of changes? All of that I think on our side, I'm, not sure it's helping or hurting honestly like it to me. It's like we can. We could move it out.

B

I mean the hope would be that you know we have fewer issues to look at because they only talk about proposals, not any single bug or every single built break I'm, not.

D

Sure it is you're done that something that can be easily kind of you know workaround by dequarius. You know on the labels, so it's just the right thing yeah. So if there would be, you know if you guys think about some stuff, that would really help. If that was all I would like. We would like to hear about that. Definitely, but the I personally pretty much run the same, but don't see much advantage of actually moving out at this moment. I think.

C

Gamers point: it's good that compiler implementation about some language designer really quite separate things and it's the difference is less clear-cut for me in certainly, it would not be clear to me where workers, because much of the feature work we do involves adding API, but it feels weird to track feature, work, separates and bugs. Let's both work for us- and it has like a single cue for a developer, so there will be disadvantages on.

D

Top of the latest enhancement, like you know, making something faster right. That's not new API! That's just like an API, so transmute in this all and that basically featural as Ellis could be could be larger. So suddenly it's kind of weird split a little bit. We need to make.

B

It more discoverable and I think the other point I wanna raise is that you know if you look at our community right now, like our github organization, it is still very much like the good old centralized version control days, whether it's there's one master of everything right and there's one. You know area everybody goes to essentially so the soon point is essentially us right like we are the ones that are arbitrating between all the different feature: requests right versus.

B

If you compare this to how the data scholar works, is that have much more of a tree organization where they where they have like? You know, there's Linus who wants the overall kernel designers they're? You know the colonel architect but like if you want to add, like a tiny sub feature to the USB driver.

B

You don't come to mind like you talk about who owns the USB driver subsystem and that allows them also to be more efficient when it comes to making decisions, because the people that make the decisions are closer to the metal and make their a faster but making those decisions. But it also allows communities to be build that are not just lying, but somebody else can say I care about something doesn't care about and they can come up with a proposal and then it's just affect you know those liners prove or disapprove of this thing.

B

But it's a it's a slightly different brick model, I'm, not convinced that we have reached the point where we need some sort of organization where we would meet that. But I think it is true that it is very hard for the community to organize themselves. I mean we have done similar things of corvex lab, but we try to build these islands of like super experimental stuff. Where some you know, team members from our team, but with a Stinnett or some other team members, collaborate on very specific issues. But we haven't done this.

B

You know in larger areas and it seems to me there's some design issues that we have seen here or design proposals where the one our team right now feels like the drive to drive. That idea- and it seems like a bad idea, but it's also not like something that we are passionate about it, like you know, spearheading like moving forward right and I think it seems that the community would be good if it would have had.

B

You know some venue where they can discuss these things in some somewhat simple as fashion versus they do in our repo. Well, we only issue, so they can water it themselves where they it's a bit of a problem, but III don't need moderation for discussion honestly, what they maybe not but I, think there's this. You know if I think there's similar to the you know how you know: homes work, but if I come to your home, I'm a guest in your home I'm not going to like clean up your fridge or like apply label.

B

It's right. It's a I mean label. Application was actually a good point yeah. It would be nice if you could at least organize your own stuff I've been, unfortunately to get a permission model like you know, we can separate those Co permissions to issue permissions, but um but that's kind of like what I'm thinking would be required in order for us to really tell the community. You know go organize yourself before it comes with a fleabag thought. I'd like today.

B

That's a bit harder to actually pull off, I mean, strictly speaking, anybody can fork or FX and can say I care about area X, and then we can help amplify them like by tweeting about them by porting to the repo or whatever, and then they could basically be. You know the I, don't know like people that care about X that we don't care about. It could have proposals, an issue there and then fall back well. It's.

D

Kinda, like you.

B

Know the only concern half of that model is that what happens if we should go like what happens if they come after like three months and then we still think like you know what it doesn't feel ready and like all.

D

Right or like it doesn't fit or something right and then and.

B

This is the nice thing. If you collaborate early that you can, you know figure out like what the principles are, that your solution has to adhere to, but it it has the dancer that it takes long all right, like I mean sometimes it's better to say you know, lady prototype something people something and then, when I think it's ready, I can make it much more compelling push for it and I can.

C

Abstract, so we also said that we talked about I'm just kind. It's.

D

Time next topic, yep.

C

D

C

To cut off, emos won't be coming for questions at the end. So yes, that's.

D

What we want so the next topic was the large community contributions, so in the post, I mentioned couple of them, so in the past or you know, last year, half a year we had signed xml and serial port coming through. Actually, we had, like you, know, orange lights in progress, but right now in drawing like you know, significant traditions like a couple of people working together and videos on directory services, I think there have been some as well. So that was you know mm-hmm, that was nice.

D

There's by the way in progress, the fastest cell stream teams award is actually helping or actually fully organ. You know working on that features, but it's a kind of larger feature. I think it was like physically some. It is one month's work and now it drags for quite longer to finish it off, which I don't mean anything bad by that by the way- and there are there actually thinks. Another ongoing thing is right now: FreeBSD port, but the kind of with the help of FreeBSD community mono at FreeBSD.

D

So if someone is interested like joining that, you know I define this once and it was actually come into driven and there's many other major examples. If you have any other program.

C

D

Your missionaries verify so anyway, so that's just a list which everyone could read from the from the issue. So it's not been super exciting I, wonder you know what people think about like you know how to best announce those things. Do we have enough of them? You know. I will be very interested in feedback and opinions of contributors section on this topic.

C

There's been other efforts like this, like um you know, usually let issues where there's a whole bunch of like check boxes right and I mean one that comes to mind is a movie dead code. There were some others and I I think they. My impression, is and be interested hear from contributors is that by and large is going really well I mean they've really been helpful to us. It's not it's! It's a huge force multiplier and there's some folks who just made really substantial contributions in in their area and I.

C

I feel like this has been really really gone. Well, so.

E

I think that today it's pretty easy to go and find issues that are up for grabs that no one has started on and are on the list of things to do, but the the actual core FX and core co. Our team might not get to the needed future. But it's not very easy to go and locate things like this, which are in progress but needs volunteers, and maybe.

C

Sheva, you know, needs more volunteers tag, that's a good! That's a good feedback down folks online feel free to you know. Let us know, or you think.

B

Like that table is inflated, alright.

E

I, don't even know if a markdown file would definitely help to centralize it, but even just a a label that you could query through github to help find this would help as well. If.

B

The reason why I suggest a markdown file is that, like, if you have a table where you see description and then a problem I could take a short summary. It may be easier for people to say oh I'm interested in XYZ, rather than going through a carry and then like trying to like figure out. What's the status of it. I like.

D

That idea, actually it.

B

Wasn't kind of.

D

Like you know, running larger projects and stuff, my question will be much more.

D

Are our contributors? Are you guys online like interested in like working more on something larger or are you run you're happy at, like you know, yeah I fix bug here back there at an API I, don't think that are super small. Sometimes it takes even a week right, but you know what are your preferences because I've seen you know some like you be, and you know a couple of contributors actually saying, like you know, hey I would like to work on something larger, like the game. Scores well mention that and.

C

Related while folks are typing, another question, in my mind, is the some contributors like you who just happy seem to be happy to just jump over and just find some new place and we contribute and then there's some folks who really specializes a certain technology like this, you know there's some people who really do link this a couple of people like tahno. We really focused on numerical staff. It was a great collaborator on the serial pause.

C

It's really just on co-op or work I wonder if there's some way we can or should kind of recognize who those experts are kind of community area contributors right recognizes people.

D

Right questions on the live chat either that or it's not.

B

Working keep in mind that there's it's slightly delayed right. I still see Dan talking, so the second needs to think five ten taken today. So the ask the question: yeah Jesus aunt.

E

Anna, while we're I'm Tanner I'm on the project system, teaming Roslin and I frequently contribute to the math and numeric sections of cortex. What.

C

Is your interest in mathematics, I'm fun by the way.

E

It's just a hobby interest of mine isn't Jannik.

D

For you know to explain it for people, like turner, is actually doing contributions to core effects kind of in his free time. It's not part of his job. So that's why we sometimes treat him more as an like. A no external contributors makes.

C

Actually a great example where we don't have deep technical expertise yeah I.

A

C

We're very responsive code in fixing things but and eric molinos, you know pretty knowledgeable, but she looks at you and there's another user, DTW user who have actual like experience in doing numerical work. You.

D

More than happy to listen lecture to these people- and it doesn't have to be that they are actually right now. You know you know sitting one floor underneath us right. You know, like jenner, so more than happy to actual listen to community experts, get some questions, so John will have to work on some larger features, but they do not find it difficult to find out the bug, though I do find difficult. Okay,.

C

One thing I mean going back to the topic of area experts and sometimes conflicted as to whether to invoke people's names, because you know like is there's a bug or something and I could really use their input on it, but they're not getting paid and I don't know if they're, if, if they wanna it's like yeah, you know if there's a test failure, that's heavily related to a change there. I don't think they actually broke it.

C

I want to ask, like you think you know what this might be, but I also don't want to take people for.

D

Granted so I don't know if you're rude, like hey I'm good, to work on.

C

That application, so some people I, can't really read whether they want that or not some people do like it. I guess.

E

Maybe in combination of a so we already have the code owners for people on corvex course hell are. Maybe we could have a community code owners where you you initially start by showing that you have expertise in the area by helping out with those types of issues and then, if you've done it enough, then you can either request personally or be report, or one of us can ask them. Would you like to be part of the.

C

Community people.

E

C

I mean and then.

E

You can do that's.

C

E

Like numerix experts, we've.

A

E

A JIT Contrave already and that's got community members as well for community members that like and know how to contribute to the JIT right.

D

E

Having that for other areas, that's.

D

Actually, that's actually good idea even into our ownership page. Having, like you know, hey community interested people or something like that right, one, more cumbersome right, so I mean by the way we have we have actually for for the owner site, set up multiple times, but we are basically tooling which sends them email. They hey. Something is new in your area.

D

So then I have to monitor the entire repo, and we actually have a couple of community members who reached out to me like teensy board and actually our and had friends and others who are actually on some of those emails as well. Are there requests? So that's something that you know these community interested people in some area might be interested in as well getting notification hey what's happening in my area,.

C

Week essentially have groups for the area owners right. So, instead of going to look at the paging, let's do you hat you know, area owner, numerix or something and those two people would get someone don't. Maybe one group includes everybody.

D

That might not be a bad idea as well, so.

B

Yeah could actually create teams. You can take the whole treatment news.

D

Are they like low? You would have to find out if it's kind of a repo, /, / organization, maintenance of those things yeah might be interesting and challenging, but yeah if it works out them. We.

B

Could just have a name prefix in the team I mean the teams are global and core.

C

Affects something yeah yeah sure it was that issue with the perfect link, sorting and I'm like at John Hannah, because he's always searching these things, but there's actually no reason why he should. You know necessarily want to have anything to.

E

C

Could do at you know, link area interested parties that would.

E

Be easier and I believe members of those groups can leave themselves they so desire. They just can't add themselves.

D

Gabaldon Alkesh limo. How is it about, like you know, adding you know, contributors to actually those groups I think they have to be member of the organization and there's like a business that are not so super nice. There.

B

Is a fundamental issue of operations that effectively in order to add anybody to a team? They have to be a member of the org. Now we could make people members of the orc and not give them access to anything. I mean that's technically possible, but it means that they all get the badge or member it so that it's at some point it becomes. You know in this thing we should work on who's actually in or number versus who's.

B

Who just happens to be the community and happens to be somebody who want you want to tag by a team. Let's see that's the problem, I mean we have. Some would get that before about that I mean like when they were here. A few months ago we raised some of the Commission concerns that we have in there they're, looking into making it more streamlined for larger organizations to be able to with their communities involved, making them team members, but yeah. So far it has been now a 1-up thing.

B

I mean we can add people to individual reposes. Why not about making important efforts? But this.

C

One really awaited Tim, Tim Tim said external. You mentioned that email system, you're sending me a great. That's your Python thing. Doesn't email external service right.

D

If the miss external is getting emails from your Python system, okay, there's no Python. It's c-sharp, but yeah yeah, listen, I'm! A solid are sexual I.

C

Didn't know there's a way for external people. How do they get on those emails? They have to send.

D

Me mail I, see asked me so yeah.

C

Okay, so that's that's worth everyone knowing I guess if you want to get emails about changes to issues in your in a particular area of this email, Carol and they'll, set you up with killed you and also accepts bright day and have enough cheers back first yeah.

D

About that, are you working on how do I.

C

D

Chips or juice, oh.

C

E

Think some of this information would also like, if you want to get added to these lists or if you want to be added to the mail list. That carol has. This would be information that be useful for us to have on the contributing page.

D

C

A yeah I honestly.

D

That the tooling is kind of a little bit taken, you know. Sometimes it doesn't work for three days. I just do they have dress erected and let's go because it's basically on the site, job that is not officially funded, unfortunately, and it needs better investment. I have actually did a project for that, but no time and that's.

E

About yeah, like.

D

You know I think we are ready, I'm ready to extend it to a few more people like whoever is kind of existing contributor and github and shalt. At least some of the contributions are more than happy.

C

One thing we'd like to make happen, but we have not had no success yet is, is trying to get more interest within. Microsoft is Microsoft of probably more github beepers than anyone else, and yet we're all inventing our own tools. So we're gonna meet again with the vs code, people right because they don't have any mail system. This, probably something of theirs like in whole. Did they.

D

Have done it now, I hope I try to actually convince jocks to invest into that area, and the first thing is like by okay, so yeah yep.

B

Yeah me like this, certainly is it concerned with subtle, be open source of that we're having it like. Github does not lead itself to something that scales very well to like thousands of people's right, I. Think I think when we talk to get up I think we said that he had like eight thousand repos or something I can across all our orgs, which is quite a bit.

C

Actually mentioned projects makes me think I we have used internally, but I don't know if we externally talked about it is some folks from time to time. Use then how we tried, like some of the others like waffle, and they all have pros and cons. I. Think these various reasons we settled on to have some of the private repos. We have there's a few of those, but there's a couple on our team.

C

They use them out, and so, if folks are interested in in that of it's acceptable to people to install that extension, we can potentially use that to manage projects with community folks as well.

B

Yeah well, I'm, talking to the contributors like I, would be curious. What you guys is cause us being the community right now. Thoughts are on made in this engine in general, like this, mainly something you would rather not use at all, which prefer get up issues or would you prefer if you would have also a meeting? That said, it's also a little bit more structure. Maybe a guy I'm asking because Miguel, for example, coming from the you know, unix/linux open source world.

B

Almost all the mono stuff was done, marinating this and still is to this day and.

B

B

I know I think the driver gets crushed.

C

I think we should use Windows next time. I am using without Mac hardware seems to be the weak point, so.

B

I think that I have to repeat what you just said, because I think you cut outer bite. We talked about that gum line, com1 again, yep yeah.

E

So, like I was saying, I believe the primary issue with when Ross and Freddie's mailing list was one. The community wasn't really asked. First, that was one feedback, and the other was that people didn't want to have to manage tracking this stuff across to three or four different places. They wanted to be able to send truly use one system. They wanted to be able to link issues and yars together, etc, and github was the answer for that.

E

There was a couple other suggestions that were closed, that had integration with github and stuff, but it had github ended up being the easiest to use, even with some of the issues that github does have so.

C

E

C

Be a depressing idea of more systems. Even that's I he'll keep. You know that make us go to another place meeting. This is 2017 salmon. Mano use mailing lists a lot, but yeah I mean I.

B

Think part of the problem of.

C

B

Grande various tools are like literally from the 80s I mean I. When I sign up to one of the on waiting lists. You get a warning that says, don't use a strong pressure because you scored in clear text on the server like that in the meeting. This tools that are as invoke on the analytic side are a bit weird.

B

But to me it's it's less about the tooling and I'm, probably use you know, exchange hosted distribution groups if we cared enough, but the question is like movies thing: may miss is the right model to begin regardless of tooling I mean I, mean there's always this problem of mating. This I think tenor. You know he didn't man who said like you, there's no history and, like I, mean there's publicly archival of those things where, like browsing, these meaningless archive is horrible. Yes, please bad, and so when you take it at the moment, you sign up.

B

You start saying you know getting email, but you don't. You have no idea spending to its I.

C

B

C

Github is probably the way to go, but it's it's the least worst right, I mean it doesn't have any good way to have threaded discussions and.

B

So I think the downside of github is and I mean that's. Why I'm still asking what may be said. Mainers have particular problems, but what I find on github is that, first of all, you cannot really delete, but if I delete it delete for everybody right, but email at my inbox, I can do whatever I want. So I can keep my own focus. So there's a threat, I don't care about, but maybe Carol cares about.

B

I can delete the threat so I can I can keep my own sanity by only looking at the stuff that I care about right, but just much harder to do on github, because you're gonna have to like do not select threads, but there's no history of like what I already look bad. Like you know, keeping keeping up to date with git up is just fundamentally hard. I think that.

E

Leaves from talking to some of the other people on the Rossum team, one of the things that we wish github had the most was an actual forum like codex had because issues aren't always the best place to discussion. Just discuss an idea and I've said this a few times. I think the only difference between us as Microsoft employees in the community is that we're able to discuss our issues with in ideas with people who know before we post it on the Internet right.

E

So that's why all the ideas we post end up looking at least decent at first and half the ideas the community post up. May not it's because they've not had the chance to discuss that and I think people think when it goes into an issue that that's an actual fleshed out idea and that's not necessarily the case a lot of times. It's just a discussion on let's flush, this out, because I have an idea, but I need to know what other people think about it.

C

She begs a question for community focuses: do we do a good enough job of keeping our discussions transparent like how often does it feel like Oh? Apparently, some discussion happened there and then Microsoft folks weigh in with the result of that, I mean you're. Never the read with we talk to each other and that's probably a good thing. So I hope there aren't many cases where there's a big discussion, and then we come back with a decision or something without really enough. Transparency.

D

We do nothing, we did that in one or two cases in last half a year, but on per because it was just unmanageable I think it was around the priority queue and something like that you know there are. There are times when actually it's good to take it offline. But it's a good question, like you know: if people feel that it's used more than you know, rare cases, I would definitely want to know I. Think to me the the.

B

Biggest shell github is that said: like did have issues alike box right there. Like things, do the number you can crack right? That's the purpose of it appreciated. I. Think that the thing we are missing is basically the structured way for discussions right and I mean the best tool that people now often have the daily left. Integrated is email like better words right I mean even internally. We most of the stuff we do is very email, even though we can meet each other.

B

Usually we still do a ton of stuff over email because it is easier to learn, express thoughts. I think I am like it's and I. Think Tim suggested is that guitar is not a replacement, because I am is like face to face. I did like if you can't do Facebook you know face to face, you do I am but I think it's it's it's the same thing you Nick, it's very it's impossible in a Nicki Church ad to to make a structure. This.

C

Other issues, I'm to be honest, I, don't go and get out anymore. One is to this ya know the place like it would help if it was integrated right there. Yes- and another is, is that because we're distributed and we've all got day, jobs we're not necessarily there at the same time, and so it becomes really disjoint I think, ideally yeah. There would be a forum like for people interested in networking or something in core effects.

C

There's a lot of things, get up to do, I mean and we've made representations before right, and we could talk to. You know, ask you for things and we we probably could have another shot at talking to the Microsoft, open source people who I guess, contract with github and see. If we can ask for things yeah.

E

I think at the very least beyond that it would be helpful if there was a way to set to indicate that this is, and this partially relies on the community doing so, but the ability to say that this is just a basic idea. We need to discuss it for first and I need to discuss it with the experts.

E

Like half of my ideas and proposals for the c-sharp language, Bo have only been accepted because I went down to Jarrett's office down the hall and talked for half an hour to flesh out the idea before I post and there needs to be a way to say, like hey I, have this basic idea: I need to discuss it with the community experts when they have some time, even if it's just one-on-one, if that's possible and then once I have an understanding of why this water would not work, then I can make an actual proposal.

E

C

It make sense in this brainstorming to have kind of office hours for particular area. Like a major like you could have a particular hour. You know four o'clock on Friday or something when the networking folks are in a gated. Channel I mean that in their office working but like if people want to have a live discussion with the DFC work on it and other community people interested I, don't know- or maybe it could just be a particular hour and get it there. Anyone in corvex control.

D

Technically speaking, we kind of have that it's very rarely used, but we had a couple of occasions when actually the discussion over github was like you know so muddy and like so unclear, and we thought that it's not probably you know, might be good idea that they actually took it offline even with the community members. Tim is actually one example of that in seaboard. He was part of the one of the LPN or no no ma one of the mssl stream proposals. Api discussion.

D

We had actually pto Lepofsky part as part of the discussion around the priority queue, or at least like you know, we chatted or fine. So when it's needed we, you know it happens, but it's yeah. It's it's rare. So.

A

D

Think something more, you know hey. You know if there's something office hours like one someone yeah well I mean maybe.

C

Maybe this is the place I'm, working and I. Don't have to say anything in this it's relevant and we could have office hours 10:00 to 11:00, Tuesday and Thursday or whatever, and anyone can join and to see what I mean I don't know like.

B

Just actually because I think the an or we lose track of the discussion first of all, bad, but also, secondly, I think it's not I, don't think it's face time. I really I really think it's it's like, if somebody did say proposes, I mean he's accusing ready time. I just went to clone at the some guy talked to me and said he doesn't want it like he's sick in his co-pays. Every always has to use these. You know constructed generic types wants to.

B

You know he has a dictionary of Topalov lon super convoluted type, so I think he wants to basically using alias with them, but still parameterize someone generic types, and you can do this in two shot. So I started a new method with the C language design team in tips. Alright, is something that I've heard and here's the three or four things that he tried to like. What are you guys and nobody.

A

E

B

In mind, we'll just ask him to say so we could do type s. You could do this. Do that and seen there before, should your type tips, but I can sort of baked that year and I think people also have to chew on this for a while, like it I, don't think you want to jump. Let me design, don't necessarily have already answer for that question. They have to think about that steep over that and like might have their thoughts, and sometimes this happens with mats as well. I mean you just publishes.

B

His current is awesome. Other reference I think for complicated design. You need you need something that I think is off line in the sense that it is, you know, face to face or I am, but you have to come up with the answer on this one. You can sleep over that think about it and then reply so I think something like either email or a forum or something that is structured or conversation. I think is the right venue to do these design things. I also think it's better to do.

B

Them are a github issues, because github issues imply they immediately like they're, like they're like like arm bumps right. There now ticking, there's like a time ticking, essentially because it's a it has a status, and people are now getting nervous, nothing's happening here, but also get some threads up flat. Like I mean like.

B

If you have like a hundred plus comments on an issue, the issue is effectively there's no more weight, you can relased is a meaningful fashion and the things some of the designs will need three four or five iterations before the design materializes into something. That's full and and I really think that we cheat a lot on our stuff, because we do this internally over. You know, because internally in meetings and then a senator said like once, we have something halfway decently designed then it's when we go to get up.

B

I think that that doesn't work, I think for the community and also doesn't really work for us. Like I, think people are also sick of like porting their market on pots, and we play to people be and, like you know, the emails from here L to github, like it think it's all work that seems to be just had a venue that is to be public I.

E

Think that's kind of one of the things where the C sharp leg repo helps. It helps separate actual work items from proposals and discussions right because the the seizure of language, though allows us it allows people to create issues, and then the Mads and Neil in particular, are pretty good about going through and saying labeling. This is discussion, proposal, etc.

E

People are allowed to discuss it. Then they're expected once the discussion has been finished, to go and create a formal proposal that has to fit the actual template, essentially with rationale, and example, usage and everything else, and.

A

E

Once that's been championed, a separate champion issue is created, you check in the spec, let for that and then you're allowed to go and work on it in the actual repo. As soon as LVM says, it's good yeah.

B

I think I mean we kind of have that and all we have it on that slash designs, repo, where the idea is that people submit pull requests with, like you know, an actual document. That is something like this Declan I I feel like the language design and for some of the you know like the the the the recent thing we had was like. How would the JIT expose hunter intrinsic so I like for for complicated areas? I think that that heavyweight process is justified.

B

I think for many of the things that our community does they're like in this interim stage, where you don't want to write a full spec. You also don't know what the issue is, yet it may take just 5 6, like iterations and yeah I, feel like this minute. Ron get that there's no story, because he either go down all the way with Mark Tom pauses for requests, or you literally submit a spec or you have these super lightweight comments.

B

They're, like the whole thing in the middle that there's a lot of stuff that is in there and on thing, has it has a good good representation um and also people just have questions. I mean like how do you deal with questions like? Do you file an issue and say I have a question like this? This whole thing doesn't exist on github at all. Oh, yes, that's why we have labor question. Maybe people can can track it themselves, it's so like it's always a bitter like and I love to ask a question like sometimes PVA.

B

E

I tend to on my own projects, prefix the issue with like bracket question or bracket proposal or bracket bug to at least help distinguish them before they can go and be tagged with the appropriate label. Since community can't label things themselves, that's.

D

Actually helpful yeah.

E

And I know that we do that in both corvex and Porsche or, to some extent with things like this is Ryu jet or this is arm, or this is so that's true too, maybe just like helping request that users do that when it happens, and we'll have to do that, for we.

D

Can actually put it in the issue, we shoot them, but eventually, but we have to here.

C

We have to do it, but I worked in databases where everything's prefixed and it to me makes it harder to read. But you don't necessarily.

E

You were the way I do. It is request that the community prefix it and then once whoever triage as it goes and absolutely remove the prefix. So.

D

It's kind of help for labeling. You know.

E

D

Community cannot change the labels themselves. It.

E

Required it would require a little bit of work on both sides, but just like with moving language design, issues to c-sharp laying out across land after the first like 100 PRS, were closed and told you need to file this over here order within hub or whatever they community kind of about the idea, and we hardly ever see language proposals on Rosslyn. Now you see them all open on the correct place, with the correct structure etc. For.

B

Most of them at least I, think processes, fine I, think every repo every technology has their own process and how to do things and like everybody, kind of expects, like a section to meet me or a city, call that I have to meet before they submit issues, and some people get offended when you tell them that they do it the wrong way, but I think that's just it's just something: the army I'm, just not sure we actually have a meaningful process for these.

B

You could continue going down the github issue path, but I, don't.

C

Know didn't see my relocations, some weather. We need a quick X API.

D

Then is actually voting for them interested anyone else on the chat can say boss. One on bands, like you know, hey I, want that to Kevin.

C

Him and I: don't I'm not paid by sent out, but there is a button if you install that and there's other extensions like the.

B

Google rotor thing that I think it's called github issue: mubarat net.

C

Or something that's a URL, you can go because they don't like rummy, because the old one still say it's a separate it's closed and the new ones kind of replayed. Yes, but it works, and this is one.

D

Click yeah, yes, Nick is gonna, knock on get-ups door very soon.

C

Yeah Nick go for it. Man.

B

Yeah I mean we don't actually find out. I did be actually mentioned, moving issues, II, don't think so. I don't think we did or Nina, but yeah I mean there's so many didn't issue, but I mean yeah, no pun intended I. Think on our side like we'll be what we should do probably is decide where that we have to separately before API lay the things we already have the notes people. Maybe things will be purpose that one that will be a one way to do it, push it down there.

C

B

And then the other one was: what did you do with questions and I mean like we can just say you know what it's it's it's get up for now, as we don't have any other alternative that is viable I mean we do have to donate Foundation forms in case anybody hasn't noticed, but that's part of the problem like it's too far away logically I mean it's like forum start.

B

D

On yet another system.

B

Basically, I like it, it's not connected, it's not part of the flow right but like there is a form I.

D

Don't see many many boats, maybe you know man and maybe Jonathan, not sure if his plus one was for that, so I would probably right now, don't don't see it as a like. You know, large community interest into moving API.

C

Just like I'm, not against it but I think it. It starts off with a minus hundred points thing because there's a cost to it and I'm trivial. You know and teaching Guerrero I already like herd cats to get all the deaths to the regularly is that late.

D

Alright, so we have ten minutes remaining. Anyone has any random question. Just ramble tall bit old, brittle stuff.

B

Is there anything that you guys feel like we should discuss the V Aaron you brought.

C

Up is why is it 25 that looks like big debt? Yes,.

D

It was later based on the basis on my team that.

C

He's under an alias today, you forgot to change his picture. Apparently.

A

D

Just read him out: man.

D

Well, questions O'regan answer, I guess so.

E

I kind of have a question or comment related to the first topic. We went over like the.

D

E

D

E

One of the one of the things that I've noticed is that I don't want to start work on a API until after it's been reviewed because I don't want to waste my own time or your guys. This time doing work, that's not going to be accepted unless it's.

A

E

On top of that, there's sometimes especially in the areas I'm interested in my changes are cross-cutting in that they also end up catching core CLR because they require intrinsic support today, although maybe with hardware transics, that will become easier. What is the recommended way to help push the proposal for and after it has been Marcus I.

C

Would tack like tag someone like me, Cowell, emo and.

A

C

And what does that go to us yeah? Obviously it just hang it like. I've, never found anyone kind of misusing tags like I get a hundred tags a day right, it's totally fine to say: hey guys, you're going this or hey, you know, hey I know, there's a head, a bunch of headed me in a queue but like I'd love to work on this and I can do that pretty soon. If you guys have chances to look at it sooner, I think that's completely reasonable. I would.

D

Do to be honest, like whoever makes me that I know the name, because you know the person worked on one or two issues at least like I'm I have no problem, helping anyone even push things forward, and that's tough yeah, because that you know sometimes things don't get the right attention. It's because there's as well lots of things. It's like! Oh this, you know I would do it. Do this and look at it like well. That doesn't really belong to core effects like really, but how do I say no to that?

D

And then you start argue and you get lost in that so yeah. That's why I kind of some of the higher value, higher quality kind of things might get lost, and that applies to bug reports that apply section to API proposals and everything. Whenever.

A

You're interested in one session, a party tactimon's ready for you well.

E

The particular case on my end was I, had the the math API, which we reviewed Tuesday, and that sat around for like a month with no update and even as a mark I had no clue where it was in the queue. What ap eyes were ahead of it? Why they might be ahead of it or.

D

So little dirty little darkest secret is that there's no queue which we set right? You know and we sometimes you know new things that actually need to get into next release. We kind of prioritize over you know, I, think that in May between April and sorry between, like you, know, January to May, actually we were able to get basically zero finding things that we had like you know, almost everything in too deep approved, or you know, reviewed as I said, like we have a backlog over summer, so we're gonna go for at once.

D

You catch up, I, believe that we're gonna be again in the mode that we can deal with everything. Yes there. We.

A

C

D

C

You, of course, can come to the.

D

Meeting so yeah any.

A

C

You turn up it's kind of hard to not look at your issue plus it to me. It's the landlord someone willing to work on something I'm pretty motivated and.

E

I guess the other thing: it's not always clear on the issues today, when a community member proposes something whether they're proposing it, because they think it's a good idea or whether they're proposing it because it's block or whether it's one of the above and they're, also willing to implement it. I know with all my API proposals, I'm almost always willing to implement it and and both in the runtime and the framework, along with all the tests and perf and everything else, but it's not always clear. So we never the case for other proposals.

D

Let's think up of people saying, like you know, hey I'm, going to implement. On the other hand, for me, if it's a good API I, don't care. If that person signs up right, it's usually like. If we approve something, then we said, like you know, hey someone wants to. You know, implement it and usually I expect that the person who file it, like oh yeah, of course, I'm you. Finally, we approve my thing here a day for last month. Obviously, wouldn't.

C

Prove it he's nothing if someone says hey I'm, you know I have time pretty soon to work on this. If you insist, let's.

D

Begin yeah we don't mind if we have like 200 approved API is we are just waiting for implementation? That's! Okay! So do you know if someone is ready to implement it doesn't change the outcome of the API review itself, but.

E

I mean like the ability to say that in so that it moves higher up in the priority. Is this Tagus.

D

Yeah that and that's true from the power to perfect.

C

E

Yeah, the other actually.

D

A trick I use that you know because I was on vacation. Unfortunately, already no outs for most summer before summer, I know that I did actually prior to the meeting, and she looked at the issues and I kind of look. Okay, these are sensors, it will be very non-controversial so kind of you know, move them higher up and then truly like you know, either unclear or more difficult kind of have them like go for direct so again, like they're, very otech, various techniques and I.

D

Much more hope that we're gonna get very soon into the point that we don't have so many, but even some of these, not that bad, like hemo, told me stories that you know more than a year ago, they actually used to have, like you know, five pages of the ready for review, which.

C

Is like we have 17 and I? Was you.

D

Know so I think that from that perspective, we're in Golden State. Of course, it's.

C

Also nice thing: we have 45 that are approved so those folks either they didn't notice, it got approved or they they don't have an interest in and.

E

That's okay, just fine and I guess the other thing is it's not always clear or sometimes there's large proposals like the hardware and critics proposal, where you know the general shape for the API, but you're not going to go and print out every single API in that list, because that's essentially as difficult as implementing it.

E

For example, the the act, the original one from Intel only covered the vectorized forms, and so of course, after discussion with Fiji and emo and Eric, and everyone else, I created the proposal for the scalar version, because that's going to be needed for its own things. But of course that's also several hundred api's not going to list them all that it's going to list the shape. What's the good way, what's the best way to indicate.

D

So I have I, haven't seen that almost at all, like I've seen and people usually do they're like reasonable thing, like you know, I've seen actually James go recently. Actually I did something like oh yeah, just gonna be like you know, 20 more functions like that and make sure it was like. Oh it's not clear which of them so I asked Mike. Can you please negative? Always not. Of course we asked him at the moment would be like okay.

D

This seems to be good and ready for review, but it would be good to kind of see the floor less right. So I think that people are doing reasonable things and we're reasonable as well. You know kind of bunch of bureaucrats right so I.

B

Mean this is I, think that this is fundamental difference between community driven stuff and internal stuff like when, but in general stuff happened, we would decide or kb1 how to intrinsics.

B

So the fashion is just half and then somebody will code up something with their devil devil basic it to the form of the whole thing, and then they will bring us a full thing and then we can say: okay, then issue is or why do you put everything on one type of is you should have one tie for a feature area or like a feedback like that, and then it would, you know, restructure that stuff, but they can mean for the most part, I mean in many cases. Puzzle is not a solid.

B

First time we took a look at. Wasn't there really had like I, don't know 20,000 eight guys like they they they didn't come to us when they were talking about well, how would we potentially expose the parser I mean they already said?

B

Okay, if siliconized it's amending model like they already had the full thing figured out at that point and I think in many cases, I think the API reduce would be better if the person submitting that already has a design thought out, because I think it is you're you're way better to explain the better to to sketch a design. Even if you don't want to sketch the whole thing, even even if Lawson would come today, we wouldn't force them to like put a markdown file, 20,000 API issue.

B

That's pointed right at that point, you find another way to do it, but like for the most part, most of the I review with Mississippi that we get are like you know, small API, small tweaks, I didn't, but most are horrendous.

D

Probably the top I have seen most.

B

Of the work that we do is not reviewing the API also the work we spend on it's like. Should we even do the feature next record is made us. The feature makes sense for the VCL it should it be somebody else's library on top, that's like 50 percent of the API, which we never do in child. They control anything.

B

It's never this like to do the feature at all, because it's not our call but like for community driven stuff like it always ends up being 50% is the bar check and the other 50% is that they actually keep it on the design. I.

D

Would just like say like we don't do it internally, because mostly those questions are kind of pre-approved. Almost so pre answered right.

B

D

Already strong denieth, like you, know, yeah like you, don't discuss that and we don't get the you know, really crazy internal questions. So just won't do this law doesn't mean that you know. Whoever is you know anytime Microsoft is privileged, but you know those people are usually kind of self filtering them. So well, okay, all right. We run out of time before.

B

We go I want to show one more thing like one more question. Brendan on Twitter found the API in web archive, and you can see the original observation comment there before we fix the comment. This is just a temporary work around the tee up about the public but hidden API, to check with an API application targeting.

B

Obviously, entropy the office lesion. If you did I know it says this- is an infrastructure API, that's not supposed to be called from your code, it's funny how, like the internet, never forgets and all your sins are recorded. What did it say before so you could figure. This is. This is what it said. This is the Highbury I say current version, says infrastructure, don't call ok, so we're working.

C

B

C

Even Isis cow, but we're gonna I wanted to know whether this is useful. Yes,.

D

That's something like that: I would actually well be probably result that or the last time I expected people who will now listen and come next time. If it's not useful, we plan to do one more in roughly one to two months and I would actually encourage everyone to go and vote on the topics or add more so that you know it's actually useful for you and if you don't get eventually any topics or any votes, we know that no one is interested in that picture.

C

Of what you're describing is basically the weekend on net the Bertrand used to do. We still do that yeah, but activation. So that's it! So are you familiar with that Victor it's summer done epilogue and it's every week you know here's all the cool libraries and conferences and features yeah.

B

Again, I want to see I'm not I have to be honest, I'm, not a fan like as in. If you go on channel 9 right now there it's like literally like three or four shows that basically the same thing, you have two people going through a PowerPoint of screenshots of postings and I'm, like I, find that, like I, mean there's already enough content. I think that does that well, I.

C

Think my horse doing it and it's nonsense.

B

Because they're more correctly we're more lower level pieces, I mean you're, not there many posts about you know list of T Berks or how how the JIT is implemented. I mean there's a few every once in a while, but it's not something we can fill in on. We.

D

Have that listing course, you are Matt for an actual story. Is that so.

C

In Japan to your question, perch and move to another team in Microsoft, Diigo Asha, but yeah.

B

No but removed to the xamarin designer team meeting. All these they're in doing shows and still doing that yeah all right all right. All right.

D

Thank you guys so again, like please vote on the topics for next time. If you're interested this time, we didn't get much more too many votes, so vote it up and.

C

Let everybody wave and smile is your comments, sir, about the different API level. Fine I, like.