Agoric SES, 5 Sep 2023

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: SES Meeting: ShadowRealm Stage 3 to 2

Description

Daniel Minor reached out today to inform the champions of ShadowRealm of their intent to present a proposal to move ShadowRealm from stage 3 to 2 until the champions have an opportunity to complete the corresponding HTML integration proposal and to give the champions an opportunity to present an update to that effect instead. Participants on the call do not plan to present an update or present a counter-proposal, though Mark as a champion of ShadowRealm would prefer the proposal remain at Stage 3 but cannot promise HTML integration work in the near term.

A

Kris “cowbert” kowal: welcome to the cess meeting.. It is august, 30, and today, on the agenda. We have our.

A

Kris “cowbert” kowal: guest, daniel miner., who is here to talk to us about shadow realm, and whether how, when? etc., to retract. That proposal based off of browser feedback.

B

Daniel minor: great. thank you. yeah.. I just wanted to follow up on this., so I believe we are meeting earlier this summer where we talked about the html integration.

B

Daniel minor:- and this is partly response to shoe from the v 8 team, reaching out to me about that.

B

Daniel minor:, so both firebox and chrome, are an agreement that we can't implement right now without further work on the html integration piece of this proposal.. So my plan is to put together a brief presentation about implementer feedback for the next plenary,, but I wanted to mention it here, first, just in case the champions would.

B

Daniel minor: prefer to present an update on on their behalf, browsers and.

B

Daniel minor: me presenting something so daniel minor:, not sure if I have too much to add. beyond that,, since since matthew on my team is person, who's been in charge of the implementation.. So I can't really speak to too much of the technical details.

B

Daniel minor:, but I think in general,.

B

Daniel minor: spider, monkey's physician is that.

B

Daniel minor:, when you either need more work in the html. integration, or the proposal could stand at stage 3 without html integration in it. with that hmo integration to be a follow up,. But I believe that chrome's position is that we should specify it all at once,. So I don't think they would support keeping this at stage 3., but I'm finding to present all of that along with some technical details, with help from matthew.

B

Daniel minor: and then we can discuss it in plenary.

B

Daniel minor: okay, yeah,- that's something I just wanted to make sure. I didn't want this to come as a surprise to anyone.. So I thought I mentioned here. and I'm quite open to.

B

Daniel minor:, having one of the champions present this instead, or collaborating or something. leave it there. thank you.

C

Mark s. miller, (mm, agoric):, so yeah, so I'll I'll just., I'm you know I'm one of the the champions.

C

Mark s. miller, (mm, agoric):, the.

C

Mark s. miller, (mm, agoric): ii appreciate that the spider monkey position.

C

Mark s. miller, (mm, agoric):, is that they want to keep it at stage 3.. I would prefer to keep it at stage. 3.

C

Mark s. miller, (mm, agoric):, I understand., let me see if I can state the the rationale behind the chrome, chrome, position.

C

Mark s. miller, (mm, agoric):, which is stage 3, implies that.

C

Mark s. miller, (mm, agoric): implementers should proceed to try to implement that. It's ready for implementation. and major implementers, of course, include the browsers. and.

C

Mark s. miller, (mm, agoric):, for the browsers cannot proceed to implement the javascript feature.

C

Mark s. miller, (mm, agoric): alone., there's the only coherent way for for javascript to implement. The feature is yeah for a browser to implement. The feature is together with html integration, and as long as that's undecided.

C

Mark s. miller, (mm, agoric):, the browsers, are blocked from proceeding.

C

mark s. miller, (mm, agoric):, so keeping it a stage. 3 is kind of an empty gesture. is that is that approximately chrome's position?? The rationale.

B

Daniel minor:, that is what I believe their position to be is.

C

Mark s. miller, (mm, agoric): okay., I can't argue with that.. I would very much like to see this proceed.. I would like to see the hkl. integration happen.

C

Mark s. miller, (mm, agoric):, but neither myself, personally, nor gorek as a company., has the.

C

Mark s. miller, (mm, agoric): resources to commit to actually.

C

Mark s. miller, (mm, agoric): doing the html. integration work.. We just wish that somebody else would.

A

Kris “cowbert” kowal:, and to that I have been in touch with charity. and.

A

Kris “cowbert” kowal:. I believe that salesforce is in conversations with.

A

Kris “cowbert” kowal: with folks to get a contract to either build out the test suite or to improve the test suite for.

A

Kris “cowbert” kowal: the proposal.. I do not know whether that,, whether that they aren't working on html integration.

A

Kris “cowbert” kowal: but yes. to mark's position,. Nobody on this call.

A

Kris “cowbert” kowal: is in a position to present an update.

B

Daniel minor: okay, thank you. I'll I'll, add something to the agenda, and make sure I leave some time for us to be able to discuss this in flannery.. But thank you for your feedback.

A

Kris “cowbert” kowal: near and dear to the module, harmony, effort, and agorick's stake in module, harmony.. One of the.

A

Kris “cowbert” kowal: can. one of the considerations that I believe matthew, your colleague daniel.

A

Kris “cowbert” kowal: brought up was that it would be difficult to decouple,, or rather to create,, that there is an extraordinary amount of lift necessary to introduce a new type of global.

A

Kris “cowbert” kowal: for, for example, the shadow realm, global.. My understanding is that the same work is necessary in order to have a global for a compartment or a global per evaluators in the terms of module, harmony, one or the other,, possibly both.

A

Kris “cowbert” kowal:, or rather one or the other,, is necessary, and the other can. The other would pre presumably stand on top of the other., so it is. it's.

A

Kris “cowbert” kowal:, which is to say,, there's nothing, nothing to say for the update regarding shadowrum retraction., but.

A

Kris “cowbert” kowal:, just so,, you know, folks, interested in evaluators and compartments are going to be interested in well, are going to the the feedback has been received, that we need to invest in making it easier to introduce new types of globals and answer questions about that.

A

Kris “cowbert” kowal: there's a.

A

Kris “cowbert” kowal:. I remember that we on this call years ago, had.

A

Kris “cowbert” kowal: an intuition that browser vendors would bulk at.

A

Kris “cowbert” kowal: having an arbitrary javascript javascript object, be proposed as a global in a particular context.

A

Kris “cowbert” kowal:. It is great now to have that substantiated from an actual from from an actual attempt to implement shadow room that that makes it clear.

B

Daniel minor: sure., I mean just., I'm glad we're a little bit just, and I don't know if this assumption applies to other browsers. but internally, in firefox. We assume things for either workers or documents., and so that means touching a large amount of code to change that assumption.. So we really appreciate having thorough test written to help make sure we don't make any mistakes while doing that kind of refactoring,, and I guess the other side of that thing is.

B

Daniel minor: proposals like this,, then daniel minor:, possibly need to involve a home expertise as well,, because I'm not sure.

B

Daniel minor: least internally inside of mozilla. if,, despite him., a p. team can really sign off, and something that could have a large amount of work for the dog teams. As well.

C

Mark s. miller, (mm, agoric):: this is this is,, let me, I wanna. I wanna raise a question.

C

Mark s. miller, (mm, agoric):, I think we're assuming that.

C

Mark s. miller, (mm, agoric):, the html integration difficulties that shadow realm face in terms of figuring out.

C

Mark s. miller, (mm, agoric): how to introduce the shadow realm global without it having without having to be a.

C

Mark s. miller, (mm, agoric): new., you know.

C

mark s. miller, (mm, agoric):, a new window,, a new frame in the in the html sense.. I think we've been assuming that we're going to face the same difficulties. and that's so.. I want to want to make that an explicit question., because I'm skeptical., I think it might be much easier to introduce.

C

Mark s. miller, (mm, agoric): department globals as new, global as as new globals.

C

Mark s. miller, (mm, agoric): and just avoid all of the html integration issues, because it's not a new realm.. It's really just a scoping mechanism within a realm.

C

Mark s. miller, (mm, agoric):, it might even help not to call it a global.

B

Daniel minor: yes,, I'm I'm trying to think this has been a topic internally, and it's I think you may be right., but I'd have to ask people that are closer to the code about this.. I know we have.

B

Daniel minor: talked about this, and I think daniel minor:. In this case it may just involve exposing a concept we already have internally, externally, in which case, yes, it would not be much additional work., but I'm I'm not very familiar with this part of the code base,. So I'd have to ask the experts.

A

Kris “cowbert” kowal: yeah,, please do.. That would be really great feedback.

A

Kris “cowbert”, kowal:, right? and and mark is right to point out that there is a major difference between evaluators and shadow rom in this regard, that there is no intention for there to be any form of html integration for either of those proposals.. There's no need for there to be internal slots in the global in that context,. Unless.

A

Kris “cowbert” kowal: there are mechanisms within the browser that are walking up the scope chain to find out more information about the context that they expect to exist on a global object.

A

Kris “cowbert” kowal:, that is not part of javascript., kris “cowbert” kowal:. If that's the case,, then it would be. This reminds me that one of the that mark is often remembers and points out that.

A

Kris “cowbert” kowal: one of the neat things about javascript. That makes it a great.

A

Kris “cowbert” kowal: language start with for for.

A

Kris “cowbert” kowal: object, level. capability,, capability-based security.

A

Kris “cowbert” kowal: reasoning is that, by dint of the coincidence that the web and javascript were designed in separate standards. bodies, they they they have that.. It has evolved into.

A

Kris “cowbert” kowal: a a clear separation of concerns between the language and the dom. or the the language, and I/o in general..

A

Kris “cowbert”, kowal: and so kris, “cowbert” kowal:, but this, the global object concern is sort of the the anchor of all of that. That makes it not quite perfect. so so let me suggest that that if we just call the compartment global, something other.

C

Mark s. miller, (mm, agoric): that well, it's a little bit difficult to call something other than global,, because it has to b, its its own value,. Its own value for this global.

C

Mark s. miller, (mm, agoric):, so I suppose we're sort of stuck with the name global. but it,. You know,, as as our shim demonstrates by emulating it,, using width.

C

Mark s. miller, (mm, agoric):, it can all be explained.

C

Mark s. miller, (mm, agoric): in the spec mark s. miller, (mm, agoric):, without really resort to any of the ceremony around the current global on the spec.

C

Kris “cowbert” kowal: yeah,, so we could, we could introduce to the specification something like the opec, scope, proxy behavior, really.

A

Kris “cowbert” kowal:, which would kris “cowbert” kowal:, terminate property access,, but not terminate necessarily a walk up the scope chain to the realm's, true global.. That might be exactly what we need in order to decouple these issues.

A

Kris “cowbert” kowal: that that would also mean.

A

Kris “cowbert” kowal:. That would also mean that we could have an almost arbitrary value. stand. There.

A

Kris “cowbert” kowal:, that's worth thinking, about. alright., well,,.

A

Kris “cowbert” kowal:. I think that we've, given you all of the feedback and and all of the additional follow-up that that we can get out of this topic.. Given the the folks present.

A

Kris “cowbert” kowal: I.

A

Daniel minor: do., you have a communication channel with kara. If you needed to follow up with him to make sure that he doesn't want to present a counter argument. okay?. So.

A

Kris “cowbert” kowal: yeah, I'll I'll get in touch with you.

A

daniel minor: sure, thank you., kris “cowbert”, kowal: alright.. Does anybody have anything that they wish to add to the agenda ad hoc?? I think we can.

C

Mark s. miller, (mm, agoric): I'd I'd like once, we go off the record. I'd like to ask a question.

A

Kris “cowbert” kowal: alright, thank you.. Let's call that assess meeting.