11 KiB
Executable File
11 KiB
Executable File
title, date, zoom_note_id, source
| title | date | zoom_note_id | source |
|---|---|---|---|
| O-E 2026 2026-10-06 21:55(GMT+6:00) | 2026-10-06 | pEKfS6rcRlWzQyGKBbqqFw | https://us01mynotes.zoom.us/pEKfS6rcRlWzQyGKBbqqFw |
O-E 2026 2026-10-06 21:55(GMT+6:00)
Transcript
- 21:55:16 · Peter Dolanjski: I don't think you should change it anyway, what you're doing, but… You know, uh, I looked this guy up. He's based here in Ottawa, in Ontario. He currently works… for, like, Ottawa… Ottawa Transit. He's a relatively new… new grad.
- 21:55:27 · russell holt: What? No kidding. Oh, you wanna hire him? Let's hire him.
- 21:55:34 · Peter Dolanjski: We can just hire him. No, seriously, like, guarantee our salary is, like, 3x what he makes right now.
- 21:55:36 · russell holt: Again.
- 21:55:40 · Peter Dolanjski: And then just tell him, you can keep doing sponsor block, we're not, you know, we'll… you can keep building that, having it, whatever. We should seriously just consider, if we're gonna discuss all this stuff with him. Like, agree on, like, we could just hire him. You know, so anyway, just a thought. Oh.
- 21:56:01 · Michal Smaga: So. Basically, the wrong guy was doing the project. It should be hiring, doing the sponsor block.
- 21:56:05 · russell holt: Thank you, Peter.
- 21:56:08 · Peter Dolanjski: Mikael, I think all your work really helped us understand what's involved here and all this. But yeah, when I look at all this, I'm just like, OK, obviously, if we can't hire them or we're not going to do that, we're going to have to work through all this.
- 21:56:21 · russell holt: Yeah.
- 21:56:23 · Peter Dolanjski: So that's great. But then I look at all this, and I'm like, we gotta get to agreement with this one dude. Like, we could just have him work here, we pay him probably a lot more than he currently makes, he's working just as a software developer.
- 21:56:33 · russell holt: Yeah, his product is good.
- 21:56:36 · Peter Dolanjski: or… yeah, he… he works on analyzing transit system stuff for the City of Ottawa. Like, guaranteed, we're gonna pay way more than he makes now. Uh, so, yeah, it's… it's an option, um, I think.
- 21:56:36 · russell holt: We like his product. We went through, just as an aside, I remember going through this discussion about Raymond Hill, the guy behind Ublock Origin.
- 21:56:54 · Peter Dolanjski: Yeah. Yeah, yeah.
- 21:56:56 · russell holt: He's pretty like. political about stuff like this and does not want to work at DuckDuckGo, so… but yeah, it's really worth thinking about, it's a good point.
- 21:57:00 · Peter Dolanjski: Totally, totally. Yeah, so… this… this feels like it might be a little bit different, but I don't know, you know, it depends on what this person's like, and so on. But anyway, I wish you could think about it, um, but I don't want to disrupt the way you're doing, but I gotta drop to another meeting.
- 21:57:10 · russell holt: Yeah. Excellent idea. Noted.
- 21:57:21 · Michal Smaga: Thanks, Peter.
- 21:57:27 · russell holt: Okay. Floor is open. Oh, see you, Francis. Yeah. Anything else? Otherwise, I'm going to start digging in here. Um, okay. All right, Nico. Let's do it. One of the things that comes to mind here, maybe you mentioned it, is if we fork the extension. and start… oh, Alex is here! Alex, did you wanna… we're running through some detailed stuff for SponsorBlock.
- 21:57:52 · Alex Martemyanov: Hey.
- 21:57:56 · russell holt: Um… What's your time frame for your day?
- 21:58:02 · Alex Martemyanov: Uh, it's 10 PM, and if we could just get a quick update on the CPM.
- 21:58:09 · Michal Smaga: Russell, I can… I can step… I can step down and stop sharing. I can hang out for, like, after the meeting, if you're free.
- 21:58:09 · Alex Martemyanov: Would be great.
- 21:58:09 · russell holt: Uh, Michael, what do you want to do? Okay, let's do that. Let's switch to Alex, your mid-mortem two, and then I'm gonna, then we'll go back into some detailed cost estimating for SponsorBlock. Okay, Alex.
- 21:58:29 · Alex Martemyanov: Yeah, uh…
- 21:58:31 · russell holt: Thank you.
- 21:58:31 · Alex Martemyanov: So, uh… So the mid-mortem, I'm sorry, it's not really filled with the data, just inside.
- 21:58:40 · russell holt: That's okay. Just tell us where you are and what you want to do.
- 21:58:41 · Alex Martemyanov: Uh, the… So generally from during the last week, we've had several fixes merged. So it's tab management, the original bugs, but not causing the total breakage, but only per tab breakage that were affected by the moving between windows. And there was the restoration, state restoration fix. And today we've started receiving the data. And then there was a speculative graveyard fix that was supposedly fixing the… You're on mute. Sorry.
- 21:59:20 · russell holt: What, what, what data? I'm sorry. Yeah. What, what data and where?
- 21:59:24 · Alex Martemyanov: Yeah, so… So generally, there was the third issue of the speculative fix that should supposedly fix the background process termination issue that was reproduced under.
- 21:59:26 · russell holt: Tell, tell us.
- 21:59:40 · Alex Martemyanov: special, uh, laboratory conditions. Yeah, and today we had, uh, we started receiving the data, basically the parameters attribution, and, uh, the second link in the chat shows that, yeah, nothing. Nothing from it, and total failure, I'd say, but it generally…
- 22:00:03 · russell holt: Wait, wait a second. I'm not sure I'm seeing what you're seeing.
- 22:00:07 · Alex Martemyanov: This is the second chat, the second link in the chat. And the one not resolved in the mid-mortem.
- 22:00:12 · russell holt: Okay.
- 22:00:17 · Alex Martemyanov: Yeah, so basically, uh, we have, uh… have not received any proofs that the fixes are fixing anything, including the speculative fix, but, uh… The pillars we are receiving just indicate very clean and regular workflows just failing. And but that's actually simplifies the further approach to the project. Uh, a bit. Although making it, uh, still unresolved and, um… Hmm. I have still no vital hypothesis that can explain it.
- 22:01:02 · russell holt: Well, it looks like reloading worked, sort of.
- 22:01:05 · Alex Martemyanov: Reloading sort of works, and restate restoration fix. I mean, other fixes should. mitigate the issue at least partially. Uh, but the main breakage still is still unresolved, and probably we won't be releasing the, uh, the speculative graveyard fix, because it's, uh, according to the data, we are not receiving those, uh, in any. In any form. Uh, so I'll probably proceed removing the fix, since it's not really affecting users, and it was kinda shaky. Uh, but there's three… three issues that may be affecting. The one was resolved in macOS 27, but we got the data from macOS 27, so it's not the case for sure. And we've got the two others in it races. Uh, WebKit bugs that… that look very similar to what we have, so there is a neat messaging loss. Uh, if the process starts connecting to the extension before the service worker is up. And it could be mitigated with the fix to the web extension code. And we discussed it with Marc already, but he argued if it's really the case that could affect all the web extension over. all the tabs. But if it's repeated, for example, it gets too fast on the repeated, uh… Condition. it may be the issue, and it's mitigated relatively easily. So one change to the web extension code that waits for the service worker to get up. And we may just experiment with it and we even can try to take it to the current macOS internal release with a kind of release exception to get released next Monday publicly and check if it… changes anything, and it shouldn't… it shouldn't break things, at least. Let's give it a try.
- 22:03:03 · russell holt: All right.
- 22:03:09 · Alex Martemyanov: And, uh… I don't know, I'll try to see… the next days coming, maybe I'll find something else. So that's the plan.
- 22:03:25 · russell holt: Sounds good. I mean. Do we think that, just side question, do we think that we have a similar issue with the…
- 22:03:30 · Alex Martemyanov: Okay.
- 22:03:37 · russell holt: YouTube ad blocking extension.
- 22:03:39 · Alex Martemyanov: Most probably. Because it's not the CPM only issue, I guess. It's the general WebKit bugs, if it is the case. Uh, it's the overall bug that affects everything.
- 22:03:44 · russell holt: Yes.
- 22:03:52 · SM: But this is an issue with, um… Content script to background messaging in the extension.
- 22:03:58 · Alex Martemyanov: Yes.
- 22:03:59 · SM: Yeah, so YouTube should be okay because it doesn't have a background, it's just a content script.
- 22:04:04 · Alex Martemyanov: Hello.
- 22:04:06 · SM: So it will. It will just run and.
- 22:04:06 · Alex Martemyanov: That's right. If, uh… Good. If it's the state restoration issue, and the script is not loading.
- 22:04:15 · Michal Smaga: Yes. Yeah, so I think it depends on, like, in which category of, like, WebKit bugs we land. Like, definitely it's not messaging dependent, so if it's, like, something messaging related, it shouldn't be affecting the Utah blocking, but if this is, like. some, like, background web extension background process, or, like, network process, like, like, dying underneath, and, like, something, like, disconnecting from the web extension. At some specific cases, it may be there, sure.
- 22:04:46 · SM: Is it… is it… for any of these issues, is it worth raising to the WebKit team? Because we've… We've had, like… Well, yeah, we've been at a couple of meetups and had, like, also direct calls with the WebKit extensions people. And… I think if we have direct feedback that we're, like, extensions are basically breaking quite badly here, then we can raise those issues with them, and they can maybe expedite fixes.
- 22:05:17 · Alex Martemyanov: Specifically, these issues that I've just mentioned, these are official WebKit radars, and they are stated as fixed, and they should be released in the upcoming macOS that's currently in beta. And if it's…
- 22:05:30 · russell holt: Really? Which one is in beta? Like the, the, uh…
- 22:05:34 · Alex Martemyanov: 27.2, I think.
- 22:05:35 · russell holt: 27.1 or whatever. And. Well, that's good.
- 22:05:41 · SM: But… but 27 is, like, as a… it takes a long time for all our user base to get there, right? Like, we can also ask for a backport. Maybe to 26.
- 22:05:54 · Alex Martemyanov: Uh, they should… I mean, the issues should be mitigated by the JS code fix if we wait, if it's really the case.
- 22:06:03 · russell holt: Yeah, thank you. I expect people to upgrade to 27 fast, because everyone wants that, um, liquid glass slider.
- 22:06:12 · Alex Martemyanov: Okay.
- 22:06:13 · russell holt: Okay.
- 22:06:16 · Alex Martemyanov: Yeah, sure. Okay, so let's try to… let's try to fix it by…
- 22:06:21 · russell holt: Okay, good.
- 22:06:26 · Alex Martemyanov: against those, uh, those bugs that are officially in WebKit, uh, Bug Tracker.
- 22:06:43 · russell holt: Excellent. Thank you, Alex.