Rust Programming Language Language Design Team, 11 Jul 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Lang Team Meeting 2019-07-11

Description

Topics covered

- never type
- coherence soundness problem
- unwinding and FFI
Minutes

https://paper.dropbox.com/doc/T-Lang-Meeting-2019.07.11--Agsvu03FSbAtkpN1VlKNgKd~Ag-hWybfJJYYpBkLF83aMAsi

A

No, it knows who I am alright cool meeting recording started. Let me share that window again. There we go so we had a. We did some pre triage here, so we were gonna, go down, I! Guess we'll go briefly through these things, the ones that are most important. This item we're gonna cover at the end red.

B

Yeah, that's that's I'm. Just guess we go through basically yeah. Okay, we'll start with that. So.

A

For async/await the current status, there was a some question around a synchronization event, but I think we've resolved that we do intend to stabilize the current behavior. If you didn't follow along you can do so. We need the basic I think at this point. Most of the blockers are down, but we want to improve the test suite and.

B

A

B

The one thing I noticed is that we have noticed that checked that it like, if you fall and async on same function, and you don't use a non-safe block- that you actually get an error. So that would be good I would I will just add it to the list.

A

I, don't know I, think most of the things are assigned at this point. So anything else to say about that.

B

And nothing, my part is.

A

Okey, let's see which one next so briefly, this type ID. We don't really have anything to add to what we said before, except that we plan to put it on some kind of list of pre RFC, interesting thoughts.

A

There's I think we did. We order everything else down here. No, so we were gonna talk about beta unwinding by default. Do we want to talk about that now or is it Marlin? Okay, I remember we said central I.

C

Don't think we did say in the triage meeting I would make the suggestion that this does urgently need to get resolved. But whenever we start talking about it, it will fill a substantial amount of available time. So we may want to see if there's any higher priority item. First, all.

A

Right so we're gonna defer that I think. Well, let's talk about the coherence or is there something else in this list stabilizers or type again, they.

D

Never type of game, King I thought.

A

We were going to do after cohere after this I.

B

Think the number type is easier to like it's maybe easier to make progress a little bit because the gern's doesn't like- and it's like shaking type theory but is like. Oh all,.

A

Right so there's two problems with never type, but the one we wanted to focus on stabilization. But one thing we wanted to focus on was we said at some point. We should try to be forwards compatible with the fact that there doesn't exist, some imple or other and I don't remember. Maybe somebody wants to talk about yeah.

E

The one that I remember was the temple of from an into to allow you to convert bang to any turn ever to any type. We can't write that today, because of we requires a more advanced version of specialization than anything a'carly huh and also.

E

Right so there was some conversation around like my sort of preferred option, for this is that we just stabilize what we have today and that since we know that any from our into in pole that takes the never type as an argument, it must be equivalent since they're all uninhabited. We can just accept them all, but obviously that's like not and like uncontroversial or like obvious language change right. So can you repeat.

A

What you just said yeah so so.

E

Any any implementation of from our into for bangs, so, for instance, I'll, actually I, should back up a little bit and explain why this is important. So the the feature compatibility hazard here right, someone could write an imple of from some specific type 40 right and that that would conflict and in fact we want people to do so for their air types right and that that would conflict with the imple of from bang for any type right.

E

The sort of trick here, which is the thing that I was suggesting, is that, since from bang, has the imple has one associated item which is a function and that function takes as an argument, an uninhabited type. You know that all implementations of from bang or for any type of must be equivalent because they're, all just the those function bodies are all unreachable and uninhabited. But obviously that's like not obvious, or maybe not. Obviously, that's, not obvious, and so I think there's like reasonable cause for objection to just doing this.

E

With with that sort of as a the plan going forward great, but so the plan, the.

A

Other plan might be specialization, like is that, are you saying we would just do great.

E

If they stabilize, if we had a format of specialization that was forward compatible with introducing this simple without breaking those existing in poles, I, don't know what that would be. I haven't I, don't think any of our existing formulations of specialization would allow I.

C

Don't think there's anything that would let us write that formulation in the language, but it's theoretically possible. We could effectively implement that in pull as a built in in the compiler yeah.

D

We definitely could why, wouldn't we, if we had, we cannot implement didn't we cannot do it.

A

I know why we can't write it today, assuming we had specialization, why wouldn't it permit this backwards compatible because.

E

I don't know what specialization is.

D

That allows that you would have I mean I mean, if you add impress you would have negative listening problems, because faith cannot rely on on the infant not being implemented in types. If we do not implement it today, then.

A

So that was the missing thing, I'd forgotten about so.

E

I I missed this part. What what? What exactly is the bit of negative reasoning that we allow the.

F

Input doesn't exist, you can write code that only compiles is only coherent because the input does not exist right.

A

So what for your own types, yeah.

F

For your own page I, don't think you can. Actually we can't, we just can't add blanket impulse. Backwards-Compatible II like I, must a and name in them. It is new to this version. You cannot add a blanket a book.

F

But I think that, like we should either we should accept.

C

F

Will not have this imple or have to have some special cases magic for it not like it's been years. It will kanae to be years. I. Think that, like the utility of this info or the utility of not having a special case are less than utility that people would be getting to bang in a time until the so.

B

There was some suggestion that we could like reserve, pretend that they will exist, but not like reserve it so that we could add it in the future, but like not make that guarantee, so that we can remove the reservation. If we wanted to I think that was up to option 2 in Eddie B's I mean.

E

I, don't know I felt that is implement, but but yeah, something that's like pretend it exists, so that you can't do the negative reasoning that, like.

A

Possible I mean that's exciting predicates. That says what are the kinds of things you allowed to know and I would be pretty happy, but it's plausible that we could say you're not allowed to know from bang.

B

It feels like reservation, is a hack, so yeah, but it's fine.

B

It's a temporary huh I mean.

E

If we're gonna do that hack, can we just do the hack to have it implement from.

B

F

A hack until we like it's not any more permanent than what you're proposing, because when already we would instead write the Apple like all.

A

The points we could throw away the hack and right and changed nothing else and we haven't affected anybody. But if we I.

F

Mean I think I think we have I just think we should stabilize bang, and whatever will do is that this whole situation is like way too much like turkey in any of the good I. Think.

E

I think we should stay Boise Bing, but I also think that, without that in bowl, bang is dramatically less useful. Yeah I think I.

A

Mean I feel actually not so bad about the the heck, in the sense that Taylor's logic, wall complex, is a way to, as this aerial said, there's like multiple routes to reasoning through that info, one through specialization and one through overlapping.

B

So another reason why I think it's a good idea is because it might be useful in other cases like. If you have a way to reserve implementations, then the Lipstein could presumably use this or we.

E

Come up with a more distinctive way of referring to which hack we're talking about. Because there are at least two.

B

E

B

Proposed in this thread, so the reservation rust see: reserve okay,.

E

So I is anybody not on board with that version of it like something that or something stronger, so.

C

Trying to understand exactly what people I get the impulse. Why is it that the info is critically important for us to provide an or for never type to be useful and I'd like to have a better understanding of what that info breaks without this hat, because.

E

Them the most common thing that you want to do with never type is say that it's something else right. So a common example is, if you have like a a result that returns a type that returns a function that returns results where the error type is bang. You want to be able to use question mark on that and have it like. Did the error type just disappear and that only works if the any error type that your function could have is convertible to from bang? Oh I see where the very type.

C

E

Nerds right so you're calling a function that returns result with an air type of bang, because it's a tray, dimple or something right. And then you want to be able to use question mark to just throw that bang away right. I, say.

F

Also like we could get pretty like not. Obviously the simple is the best way to do that, but we could also like have we can implement error for bang and then, like that's sort of famous any errors in library right, you'll respond from yesterday for your own in cool, so.

E

That's not actually enough and the reason and I and I can go into the specifics of why that doesn't work out, but but basically a lot of the types that you're talking about here like or wind up in generic context, where they're working with a type that is like specifically something that can be converted into a stood I/o error or something like that or a type that can be converted into you, know some other specific air type. When.

F

I but I think that the people who write those library, whose letters absolutely responsible for adding the front Google and it would just be another thing that they have to do, and that is another possibility. So.

G

That's what they're doing today with yeah yeah infallible right, is that the.

B

Way, people do that today. So with respect to the like pretend reservation thing, the main question I have a permit.

B

Maybe I will do that. Maybe what do you think you could do that I? Think.

A

Well of her already real answer.

B

A

Think I could mentor someone do that too I. Don't it's either really easy words hard. It gets probably really easy. Sorry, the two options you know there could be me.

G

Do any of the hacks that are discussed have implications on being able to turn convert infallible into bang or is.

F

G

Gonna be fine, no matter what we do well.

F

G

F

It's a choice to make it like it's not possible to implement from bang for any type. Now, if we say that, even though the info does not exist, you cannot rely on knowing that doesn't exist, I mean so.

E

So, actually, actually Scott. That's a good thing to point out is that technically today, if somebody did have this, if somebody was doing this negative reasoning to see that core convert infallible with oh, but they can't because core convert infallible is defined inside the standard library, and so they can't have that negative reasoning about it, because somebody might add that from a mental I, never I take it back with.

A

Oats, maybe we should clarify, because boats raised to pointed that, it's not what I'm I should not meet my expectations. What what this middle reservation option is? I was assuming people would still be able to write for their own types, but.

F

They're, not you don't not if they have to assume that this not a fancy backward before competitive, like is. This were compatible with adding from like I wrote from bang for my effort, then that's the kind of code I'll be broken by adding from bang for ki, accept modulus, visualization, cetera well,.

A

I was assuming that you would be allowed to write it for your own types, but you're not allowed to rely that other types. The other types they're.

F

Your own types, even.

A

Don't have it and you have you sort of can't do this pattern, I guess this pattern, you wouldn't be able to do so. Scott.

E

Yeah I, don't know, saw but I just wrote some code about the core convert and Val infallible case. Where, like like you, you can do the same thing with court convert and rot infallible today. So if anyone is doing that and then we made core convert infallible equivalent to bang, we would be breaking that code. I, don't know how many people have actually done negative reasoning based on the fact that, like yeah.

F

E

F

Yeah I also think that this is like a really impossible situation. You're.

A

Not allowed to write down well today, are you I, don't think you are because why not? Don't you don't own any of the you can't do negative reasoning on types you don't on yeah.

F

A

E

But but but you do own the type of above here. Let me comment this out. So it's less confusing you you do own fan. Foo is the thing you're doing negative reasoning on and you can reason.

A

E

A

You can do that. I didn't understand what you're saying, but it was someone proposing that I don't even know any foul is acting.

E

Like the like, from stir in bulls and stuff like that that are in foul returned now, it's.

F

A library to find bang okay.

A

B

E

B

One couldn't.

E

Stable that thing, so we made this thing instead, so but I mean all this stuff is just like.

F

It's just like whole thing is really frustrating just because you we're trying to make the most perfect possible like where everything is written for you, using this magic and arrogant, Bulls thing, and it's not working instead of being like okay, that doesn't work.

F

What is sort of the like thing that does work that, like maybe people, have to write a from banging pool for the error if they want to convert a bang into their error type like instead, we're like talking about adding hacks and like delaying everything right now, it's been years, and it just feels like the alternative, is the absolute worst case. I agree with I hear you.

C

One of the concerns like as I understand it. There is if people write the from imple themselves and then we figure out a generic way to say you can always convert Bank to anything. Do those two things conflict yeah way to make them not conflict? Then we could go ahead and release now, while we sort out how to make that generic implementation, but if we're not confident.

D

Negative reasoning: we need to open the negative reasoning, because otherwise people could do things if such types, for example, if they do this I.

E

Then give it also on this negative reasoning, isn't actually hard to arrive at after. Like writing some code that looks sort of like this, like.

E

F

E

Arrow type I can definitely see someone writing something that looks like this. If.

F

We're only blocking the negative reasoning and you can still implement from bang for your own types and I. Think that's fine, um I, don't think this, like I'm an optimist whatever having a specialization I mean.

E

How do you feel a lot of us just like hacking like like forcing that imple to exist? I also think that's.

F

Fine I think they I think I.

E

F

E

My preferred route, because this seems like something that we've wanted to add for so long. We have like that. The justification for how to write it in a library form is so difficult and, like so.

F

It would rather like have this be stable with nothing then spend another six months, arguing about whether or not it's appropriate to add that to the language like.

E

So says anybody I know I think that's roll! You mentioned that. Maybe you didn't like that. Does anybody else or central? Do you want to talk a little bit more like about your thoughts around that.

B

I'm not sure I feel like that.

B

E

You wanna talk with me off and about that, and we can decide whether that's something that we're pushing on, but I think we're all on board for the like impeding negative reasoning back right.

E

B

A

Think it's a good idea to investigate the negative reasoning but I think it's a good idea to try to bottom out this boat, because I also so like. At the end of the day. If we had a special casing balloon but though, like.

F

Prom and Bing are already like magic like from, is already and got using sugar like things right.

A

That's what I mean they're already know it's the language right, no weird.

B

Magic code word mystery, so I think, oh that you're pretty busy. So maybe.

B

D

Do so you want to try to get up a qualify for the embedding yeah yeah yeah.

A

That's fine I'm, happy with that yeah, but I yeah, okay, I'm happy with that, but I feel like I would probably be happier with the funding. So I'm wondering the last question I'm asked is: what would it take to make us feel sufficiently confident that we would have some like is a route to getting this central that we, the only route, is that we have a stable path or we feel like there's something with sufficient probability that we'll have a path to justify it. Non-Food.

B

I, so the coherence problem makes me like specialization like that. We can do it some. We know how to do this, not be adult. So it's hard for me, too sure.

E

It's on it sounds like very far away from a like concrete rationalization of specialization that would allow this right.

F

Maybe what you're gonna be able to specialize on from that was my understanding of the.

E

B

E

There's also I mean there's also good, there's also my like magic like if all the associated items or functions which are uninhabited or something like which take uninhabited arguments like right.

F

E

Yeah you're a million ways so yeah. That would be.

A

Well, if they're negative reasoning is easier, we can start with that. I guess: I agree with coats. The delay is the worst part.

C

We elicit test involves an imple of from bang for a specific type or similar, and or for that matter, for a tight, a tray like everything that implements a trait and just make sure that test passes with whatever hack we have you.

B

Write down the thing you want, do you mean like in the BR address I'm.

C

Saying yeah in the PR can we add a test case to the compiler for info for specific type or similar? You.

B

Know that sounds good. Can you just write a dump, so we got right. Thank you.

E

As a passing note, I do want to mention, but though, the the first time we tried to stabilize this, it actually got called off because and only because it turned out that the libs team was gonna, recommend against it because of the whole box. Stein, error, conversion failing.

F

A

F

We could have just implemented error on the light.

B

E

That I was that it was a bug that got fixed. It was an issue with it tried to coerce it too early, and so you were passing a nun. Sized value to box knew.

B

About the other problem, but I do.

B

A

Be nice to a team.

A

All right, so we have this other problem that we've been putting off for a while until I wanted to talk about it, coherence being bypassed Ariel. Do you want to sort of show the problem to start.

A

Do you want me to do it.

A

Right I mean the the short version is that we can have we. We have a rule. We say that din Phu Phu and we have a rule that sort of prevents you from writing very simple impulse like this. However, that rule is not especially smart and you can create more complex impulse that will kind of bypass it, for example this and in fact we're kind of relying on this and the compiler I found on there anyway do some more, but it's annoying so this would be like now. The problem is this: in pull.

A

If T is didn't foo, the simple applies, so that's not great, because now we have two definitions: either have the built-in in foo foo, or we have this simple like two ways to.

E

What code actually gets called yesterday, if you call a method on the trade object, I.

A

Think the trade object I think what happens as we go through the vtable dispatch, because, by the.

B

Time we're on amorphize.

D

Diploma that you can have an associate type and then you can have very easy and sadness with that right.

B

You can basically encode trust me yeah.

D

Yeah yeah, yes,.

B

And there wasn't there, there is a small in studying coats like a smaller version of it. If you want to see yeah.

A

That's good. There was another issue somewhat related to this and maybe was even the same.

A

That we ran into before.

A

I'll go look at it in the screen, I'm, not sharing that everybody. um It was by.

A

Who filed it anyway, I'm gonna look for some background, but I mean we kept that I. Remember we encountered a similar. She won't be toyed with different rules to rule out overlapping animals, so.

B

So the idea I had is that we would probably propagate technical things, but I think found ways to make that not work, but basically I wanted to blame I. So.

D

So, basically figure it found out all the reasons why we do want to have coherence.

D

Why we don't want to give up on.

C

D

Because you could please I can't think of ever example. Every way to buy coherence at all that looks like the products. I can come up with some way to make it and sound.

B

You want to walk through.

A

An example or two of those just to give us a flavor I, don't.

D

Know them exactly that can get a bit complicated, but.

A

This was the other issue that was related, I, think.

E

Suma bleah someone like checked what happens if we just disallow this and it doesn't like a bunch of stuff breaks right. You just disallow this kind of imple, correct, disallow I mean correctly realize that this is an overlap with imple. The trade object, I think what we were well.

A

I, don't know I, don't really understand. Well, we I think what we did. The the solution we were pursuing for this other issue. Problems, I, think, is the same, although I haven't he's. Basically, the same was to make this trade not objects. If there's do.

B

We have any like movement which check in weather no make.

E

No sorry make the fish make a trade object safe. If there's an impotent would apply to its dine, make it not objects. No.

F

D

F

It not object safe.

D

F

And I would have to like that's also fine, because that's the existence increase, because it would have to be a a blanket in valoran and before the like, it would have been orphan in ela. Creates that.

A

Said it's interesting. We we said we had future compatibility. Warnings in place are on this issue, but maybe we don't or maybe they're not sufficient to cover this.

B

So there you don't get any.

C

B

That case, in that specific case, you could get around it with any concentrates on by propagating on.

B

But I have found wave moments and they're a whole lot more complicated.

A

D

Basically make the object candidate impel have a well close that required. No other input applies so.

A

You kind of you're saying here: Ariel is basically like abandoning gay, essentially there's two varieties of you can write an input by hand for your tintype if you like otherwise like so that's how I understand that you could write the simple by hand if you like or indirectly and you're allowed to do that. But if you do it, then you don't get the automatic involve. Essentially it.

D

Only applies if the if no other impelled can apply can apply. That's.

B

Pretty ok, so I talked about.

A

How to do it, though, so.

B

I am I also live in detail. The able fire to gain, but so the one thing I think would be good, is that we like to test out intuitions, since there seem to be different solutions that are workable and find out how much we break and how much it would affect lines and stuff so because otherwise, it's so hard to decide. But now, which is better.

B

A

I agree: we should do some testing.

A

But I it seems like the last one.

A

At least it seems to have the relatively minimal potential for breakage, but we have to like write it out more a little more clearly what it means around more.

F

Complex traits I, like the last one I, think that the thing that are what a couple years ago about me, yeah, yeah, there's from there it's.

A

Kind of annoying me that we can, there have been times I, wanted to write this simple in any case, usually as a way to make the treat objects safe.

B

Do you think this interact would be rebalancing currents, I.

D

Don't think so doesn't change any of the it doesn't change any of the words other traits.

A

One thing I might apply, am I am I, correct. I. Think I could correct to say that if you had an Impala in pull of foo for didn't you for all possible instantiations or even not for all possible, actually, then in a sense that that could make the treat object safe, even if it wouldn't otherwise be like this might replace the wear. Self.Size tack, the idea would be essentially I write.

A

It I write a trade to know if it's not an object, safe things, then I write an in pull of that trade in terms of the dynamic version of itself, but I can only use the UM it would require us to like this.

B

B

F

It does, but it's only semi relevant.

A

The pr one I don't think solve this problem. Yes,.

F

A

Does kind of do some of the work that I wanted you if we solve this problem.

F

Yeah it just worth considering just been rebasing that PR today.

A

I think it passes test no.

F

But I think um well I think that what you're describing where you could not have to wear self sighs things seems obviously awesome, but that's narrowing to like what you want to dispatch the like sabi, safe methods, or at least the probably the unprovided methods, at least to the me table. It's.

A

Kind of getting rid of the idea of like the other route has always been that we get rid of the idea of object safety and we instead sort of say that you.

F

Can have a trial you get the like call the vtable methods, if you are providing the employer self, like that's the thing we would have to defend right.

A

That's the thing you have to add back, or else you can't do it that you have to have some. It makes the language you could be done as a future step. I, don't know if it's a good idea or not, but I'm, not crazy about the worst self sighs time. You basically have another kind of call. You know it's not a right now, we're out the vtable method, sort of through a special magic imple. This would be that you want to route them through a special there's. Just like their like an inherent method.

A

The compiler knows how to generate them. Yeah.

F

I mean or even just could be heading to the the RCS for having an API, perhaps a new table the devii table could like you could have some light. Reading function, players for these methods. Out of that.

A

Yeah I guess that would be like to a low level version. Yes, yeah.

F

It would not like yeah so most use cases, it would not be pleasant after then, it migrated net call the net.

D

Having an implied does is that you can't you want to have the blue sync loose and stuff you just belittle it afford any title. You don't good point.

B

So one way around that might be like funded by balance.

B

You could say time.

B

A

This is this is so I. We clearly wouldn't want to go this far in one step anyway solved on this whole, but it is interesting.

B

F

Yeah being a believer, moderate safety would be very nice, so.

A

Well, what I like about this, if we did go this way, is that the main downside of getting rid of objects safety when we first considered it was that you had to rent these impulse Ryan.

F

A

Instead, just now.

F

A

Don't have to, but you have the.

F

A

F

Means that it we often right there for you, but you can write it exactly.

G

A

Before that was a real drag.

B

Is also concept that many people find difficult to understand. Yeah.

F

That's my to be nice to get rid of it as an age. Well,.

G

I think taught me a lot to check for and a bunch of thinking it's very.

F

Leaky, it's just like here's, a bunch of random rules that.

G

People do accidents a lot. The.

A

Problem is those rules don't go away yeah under any of these versions like even in this version, it's gonna the rules, just sort of move around then they're like what can you do, but I still think it's might be an improvement. It.

C

Looked like you were trying to say something and you were maybe on mute and.

C

We're still not hearing you or I'm, not.

A

E

Alright, it seems like sometimes zoom just decides that I don't get to be hurt. I was just gonna, say, I kind of want to push back against the idea that we want to make fewer things object safe or that we like would get rid of object safety somehow and make it always manual. I feel like there's already enough sort of punishing that we do have rust users trying to use a dynamic dispatch and I feel like adding more things to that list would be bad.

F

Simply not what I mean I mean like okay, ideal like we could just like, if all the trace in center library or somehow obviously would be like videos like a flare yeah. As Nico pointed out, you don't distance moves around the problem, is you still have to right? Yeah.

E

I mean you, you were there at all hands when I tried to scheme out and was pushing for trying to make like generic methods, object, safe and it all kind of fell apart yeah it would be a deal that.

F

Would be, what would be awesome is if 50, because everything not always.

B

That going to work another interesting, it didn't work, it.

A

C

If I might ask a weird general question here, is there a like clearly documented version of this test case that includes a careful walkthrough for people who don't like trivially, translate type system rules in their head? So.

B

There are I think the the simplest version is probably the one I wrote and there are tons of different versions. I can try to. Thank you. Sorry. Sorry, the.

C

Simplest version is what one that you wrote. Sorry.

F

Paper document with the builds of a method called transmute, which is zr1 also, my name is Sarah.

C

I'm reading that example, and finding myself wishing that it had about twice as many comments as lines of code. Yes, the battle here so.

B

C

Not asking for an explanation in this call I'm, just making me out like everybody, was completely following along here am I the only person who's just baffled by this example. I think.

B

We, some of them us, have spent a bit more time. Thinking about it, I mean.

F

Like at least three or four minutes before I started speaking staring at the example trying to understand how it worked.

B

C

Does feel like we've had a couple of things in a row that involve type system issues around dying and I'm, wondering it to what extent we might need to step back and look at a like. What is the coherent tight system around dying because it does feel like we're patching it incrementally.

A

You know we're to the other issues you're thinking of I there's one listed here, but I would like to have a full list.

C

There's the one listed here and I it may be the same one but I seem to recall seeing another case that was like here we can implement transmit.

A

Be good to get a list, I think I think they all hinge. Oh I want to go review this other issue. I guess we could do it now, but they tend to hinge around exactly this sort of interaction, so it might be that indeed we can sort of solve them in a fell swoop.

B

But so pursue regarding regarding trying out some different options and and it's being what falls out in terms of testing, we sort of need a plan for doing this. Someone who is familiar with the parts of the entire business and strikes and he's out I suppose that's like Nico scary, max yeah,.

A

Skill excel might be into this or ariel depending on which time Ariel has. We could certainly like certainly ping and discuss them. I think skelux, I might be a good thought, especially.

A

D

I think it would be a good idea to have some to have someone for the time this try to figure out how this feature fair, Chuck, so I'll at least even if I planned it I like to talk with them about it. Yeah.

A

My main concern that the last one is gonna be something around. You know you're sort of having to evaluate whether another envelope lies yeah.

D

A

The assuming that the didn't through food rule doesn't isn't there like it's like a changing set of impulse that we're throwing right, which.

C

B

Can do that? It.

A

Actually seems fine, but it's just wanna make sure we all read. I, don't know using hot water. Would.

B

It be, would it be any use like talk to Richard, Eisenberg and like it's that me were from hospital, is.

A

Maybe I mean there's a lot of context to supply I think we don't think we're there yet, since you should probably investigate first her but like you should at least doing some of the groundwork yeah, let's ping skill exam tomorrow or later today, and see what they, what they think I think what needs to be done right now is to write up in more detail. What's been said here,.

B

Right I think so Josh.

B

It took a lot it some time for me to understand the example as well, but even I, like I'm at the point where I'm.

B

C

That sounds like a conversation. We should have out-of-band I just like to get a better idea, but where is the actual problem here? I see the result like to see where, along the way, something went horribly wrong right.

B

So yeah I'm writing up some stuff. You can leave this from eco, maybe yeah.

A

I think I could try to that interesting after next to my next meeting.

D

Me do it alright,.

B

Yeah I was more trying to write up the alternatives and trying to sketch them out a little bit like.

B

A

Central to writer- and they elaborated example, and why is your name in John Fox? We are not central.

A

How come you haven't even.

A

A

B

Should get it legally changed angry I.

A

Wouldn't objective, at least.

C

So do does anybody think that, in their remaining.

D

Eight minutes.

C

We might be able to reasonably at least make a little forward front progress on the unwinding through FFI question. I would.

A

At least appreciate a summary: if anyone has one of what what is that I'd.

C

Be happy to try to give as neutral a summary is that I, as I can with the disclaimer that I'm also one of the involved parties, but I will try to summarize as neutrally as I can.

C

There is an issue where certain C libraries have error handlers that are effectively. We call abort. If you return from this error handler and the expectation was always Oh print some messages, you can either print some messages and then abort or you can long jump up past the library code back into your own code and that's how you get to keep going if they're sales.

C

One of the examples of this is loop jpg, for example, where, if you have an error, parsing a jpg, then you get to long jump out of JPEG in order to not crush your whole save browser window because you failed to parse an image. This is a lovely historical artifact of see, error handling it that we said we have to cope with and in rust people handled this by using panic as effectively long jump, they would do a panic inside their error handler. Allow that to unwind BC code.

C

In live JPEG and then catch that panic in the rust code that called live JPEG and they didn't make any assumptions that live. Jpeg was still usable in that context or that it didn't have a bunch of screw age inside its internal state, but then they could unwind it. That does require you to carefully build, live JPEG with unwind it hooks and similar with the right compiler options, but it can be made to work. The problem with this is that LLVM has a flag, no unlined.

C

It, which is I promise not to unwind past dysfunction, and we currently declare all extern C functions with that flag, which means they can be Mis optimized as assuming that they will never have a panic propagate past them. At one point, we actually tried to fix this by making it so that, if you unlined past a an extern c function, we would abort rather than unwinding, and this ran headlong into this broke. People likely web users of jim live jpeg.

C

Everybody involved understands that there was undefined behavior here and that there is a problem and that the last time people tried to push this. It got reverted, but there was a feeling, like nobody made any forward progress on it. So this time around that we effectively said, let's revert it again, but let's put a 12-week deadline on making some kind of forward progress to actually solving the problem, and if we don't make any forward progress, then we'll just declare it as we're, not reverting it again and deal with it fix it.

C

Nobody set forward to fix the problem so that had the intended effect, people stepped forward to try to fix the problem and to write a real RFC. That is here's how to make a function not get marked as no unlined it. Here's how to deal with panics that has currently got caught up in a fair bit of detail about. Well, what does it mean to unwind? What about different panic runs and what about exceptions?

C

How does this interact with C++ exceptions and running destructors on the way out, if Emily, making the gold armor-plated solution that solves every use case, and it has gotten somewhat distracted from the problem of? We need to not put no unwind on these functions or we have undefined behavior so we're trying to get that back on track. There's a lot of discussion of what is the minimum needed to solve the problem and fix the undefined behavior, but in the meantime the deadline came up and there is some desire to just say: well the deadline passed.

C

So, let's reapply the change and or rather the change was only ever done and reverted in beta, and so, if do nothing, then we go back to aborting. If you try to unwind through FFI and existing code breaks, and if we merge this 6 to 5 in beta, then we defer it another six week release and allow time for the RFC to find its way through the process and figure out what it's doing. I personally would recommend that we merge this because forward progress is being made.

C

It would actually be my preference to merge it into trunk or Mia stump. You know that lets reapply, the other change of when the RFC is sorted out, but I'm, assuming that there is not unanimity on that.

C

B

So first emerging emerging from from my perspective. As long as we haven't resolved this, we should either merge it on like merged here on theta, or we should merge it on master. You have to do okay,.

A

That's good BB, at least you know, as you would prefer, that all three channels behave the same self, correct central, well,.

B

So the idea was that we would like have some pressure to get to this mission by not adding you know monster but I think. Don't think that it's nothing. So we can just merge it on mustard and you know I, don't.

C

That I think the pressure had the intended effect of. We need to actually make forward progress and I think now that there is a live RFC that is being actively worked on then until that gets dropped. Unless that gets dropped on the floor, I think we have forward progress going on and the pressure is now more generating stress in trying to actually work on the RFC. So.

B

So maybe them you know, probably not.

B

My larger point was just that, like until we have consensus in the team we should make like, we have to find they're going to do negative work. Basically I.

A

Tend to agree with that, I definitely think we should merge this PR I. Think that's kind of a no-brainer I also would be happy to revert bethe very merged semester, but I do very much want this or I see my partner's right.

C

And the one comment I would make on that is: there's been some discussion on five: eight, seven, nine four, which is kind of the the issue that the RSC is trying to address and I. Think there is some thought.

C

Could we just add the simplest possible attribute that turns off lob M's, no unwind on a function and then follow the whole? How does panic interact with other things later and for now leave it with if you're doing this you'd better know what you're doing with respect to building your C code, with the ability to be unwound, some undefined behavior, but I'm wondering if that might make people a little bit less stressed about the fact that we are kicking this down the road another six weeks. It's.

A

Kind of assumed that's what we were gonna, add I kind.

C

Of hoped that was working, we're going to add and then I go look at the RFC it's like. Why is this longer than a page.

B

But so so so, what's the idea around for a compatibility with back-end, they actually have to not tiny ourselves too much to write.

A

Probably the general consensus: we should merge the PR yeah.

F

Merge it into trunk master to.

B

So I think there's going.

F

To be a solution, and we should do everything like yeah we're not breaking like we've decided to break people own. We should just do that. Yeah.

B

F

In their land behind paper or letters, they like working on a proper solution and everything will go well and whatever I.

C

Won't be sure to talk with the RFC folks and say it feels like this has grown a little bit. Could we try to do the simplest thing that addresses the undefined behavior and then address the larger solution? I'll try to get that back on track a little.

B

So I read the arcing in, but I would like to involve Alex these impressions. From my perspective, I think the wrapping with C solution, like broccoli programs, is us like a workable solution in the entrance. So personally, I would like to change the behavior right.

A

I really do have to get off that another cause. I'm just gonna end it don't you all should discuss anything. We're discussing discord, Joe thanks. Everybody.