k

@herrmanndigital Klaviyo only sends the email address, nothing else

Meta is now providing match scores for audiences. Interesting when using the segment from Klaviyo the match score is poor vs just uploading the audience manually.

@trq212 github.com/anthropics/cla… and github.com/anthropics/cla…

we're doing a lot more of this, hunting down some of the most annoying bugs in Claude Code let me know if you have any white whales

@bcherny @lydiahallie @felixrieseberg @_catwu Any way to fix this? github.com/anthropics/cla… PreToolUse never opened the diff viewer, but this stopped working in 2.1.78. Having the diff viewer constantly stealing focus makes the extension unusuable.

@trq212 @ns123abc might taking a look at github.com/anthropics/cla…? been broken for 3 weeks now

unblocked (you and others) it honestly just sucks to be getting tagged like this when you're literally working on a Saturday trying to debug things for users- I've done so many calls this week trying to understand peoples usage the post itself wasnt bad but it leads to a lot of people piling on and feels personal but feedback taken, blocks are too broad given my role

WTF?!

@bcherny Would be great if you could take a look @https://github.com/anthropics/claude-code/issues/36286 - PermissionDecision hasn't been working since 2.1.77 (VS Code Extension)

Today was a good day

@norbert_tech @toonverwerft yeah, you need an event loop

@toonverwerft @knrd but doesn't that mean that code that needs to make http requests needs to run inside of an event loop? Its for telemetry library, I doubt anyone would like to wrap their code with my code

Recently I have been working on a php telemetry library that is fully compatible with open telemetry protocol and does not use global state/singletons. I'm almost done with first draft, but of course I had to step into another rabbit hole... Async http client.

@norbert_tech ah, gotcha, missed/misread that part

@knrd the problem is contract itself, not the implementation - once I figure out how to approach the contract problem, finding/building implementations would be a implementation detail 😅

@norbert_tech Using the amphp guzzle adapter an option?

There is something tempting about owning all those contracts and having control over everything, but on the other hand, it adds work - not much, as this would be 2 interfaces and 3 implementations (Symfony/Guzzle/curl) but still. What would you do? 🤔

@ArijanJanes 50

Made a list with 50+ listicle funnels Designs and links inside Figma. Study these and apply the same principles to your funnel. If you want the link: Comment "50" and it'll be auto-sent.

@VotrubaT @markusstaab @terdelyi I have DTOs that all have the same read-only property (entity id). In PHP 8.3, I couldn't just use constructor property promotion to satisfy the interface.

@knrd @markusstaab @terdelyi How do you use it in practice?

Which version do you prefer? A. Promoted property with getter B. Property hook with PHP 7.4-like constructor assign

@VotrubaT @markusstaab @terdelyi biggest thing for me is being able to specify properties on an interface.

@markusstaab @terdelyi 💯 Couldn't agree more Readonly solved mutability for good. One word, that's it. No idea what problem hooks supposed to solve and improve for developers 🤔

@eric_seufert Seeing as one of the companies mentioned is from Germany, I'm guessing complain to the EU?

Asking platforms to make their auction logic transparent is an interesting idea, but I'm not sure what value it would bring to the ecosystem or why platforms would support it. The types of auctions implemented by the largest ad platforms are broadly known to advertisers, even if not explicitly stated. Meta talks about the "total value" approach used in its auction, and I've seen Meta employees call it a modified VCG auction at conferences, although Meta doesn't use the phrase "second price" anywhere in its auction documentation. Google's transition of AdSense to a first-price auction in 2021 was documented. When these platforms modify their auction logic, they bring other factors into consideration -- action probability, "quality" scores, relevance, etc. These modifications are generally plainly stated and at least conceptually / qualitatively made clear to advertisers. I don't see how quantifying the weights of these modifications or publishing their logic would enhance advertiser efficiency. And doing so would obviously expose these platforms to risk: any time they made a change, they'd need to highlight it, and even publishing pseudo-code for the underlying logic could leak proprietary data. My question to any advertiser advocating for this information to be made public is: what would you do with it?

@binghott @IstvanicMarin Generally I would agree with not manually fiddling with the campaign, but even Meta suggest there's times when this makes sense.

@IstvanicMarin Mmmm, if you're optimizing for your desired goal, why would the system be spending on bad performing ads and getting it wrong? Why would frequency matter?

Advertisers, if you see individual Meta ads with high CPA's and/or low ROAS, do you manually turn them off or leave them on?

@dtcpages Listicle

Listicle landing pages still print money for the biggest Shopify brands. If you’re an ecom founder and you’re not testing them, you’re mising big. I grabbed 55 of the highest-converting listicle landers running on Meta RIGHT NOW. Framed, sliced, and labeled in a FREE Figma file. Supplements. Skincare. Apparel. Pets. And more. All ready to swipe. Want it? 1. Like this 2. Comment “listicle” below 3. Follow so I can DM the link.

@brendt_gd @phpstorm PhpStorm strips generics and replaces them with the interface/implementation when moving/refactoring classes. Have dozens of commits that are just fixing PhpStorm fucking up my docblocks.

No promises we'll be able to fix everything but… We really want to improve generics support in @phpstorm, which issue stands out to you that should definitely be fixed?

@MrYoniLevy We're also not in any of categories that are affected by the ongoing restrictions.

@MrYoniLevy We're suddenly seeing "Upcoming restrictions on custom conversions" in EM, however nothing is actually listed under impacted conversions/review custom conversions. We're also not sending any sensitive info. Is this an error? If not, how can we request a review? Thanks

@MattiSchroder @jordanmenard In addition to what @IstvanicMarin mentioned, there's also fewer creative enhancements available when using existing posts. We used to differentiate between scaling/testing and had mixed results. This year we changed our approach and the difference has been day and night.

I'll be the first to admit they I disagree with that method, but also that you're most likely a WAY better media buyer than I am. So not here to discuss that! But I'm curious to hear your thoughts about WHY this is better based on how the auction works etc. What’s the logic, if you can call it that? My concern with “it just works better” is that it’s really hard to compare one method vs another as we’ll never know how the performance would have been launching the same ads, at the same time, etc., using other methods. So that’s why I like to resort to arguments about how Meta works, not just what we observe about something we can't really split test, if that makes sense. Here to learn! 🙋🏻‍♂️

Every savage media buyer I know tests creative with abo and scales it where it lies. Everyone has their own twist or style from that foundation , but functionally it’s the same thing. Every. Single. Savage media buyer I know. Does it this way. I’m sure it’s a coincidence.

@IstvanicMarin also doesn't help that some useful enhancements aren't available when using existing posts (eg. text overlay)

People who are against isolated testing campaigns DON'T understand that I'm NOT turning off ads that are good performers when moving them to scaling campaign I have multiple ad accounts where the testing campaign is actually the highest spending campaign in the ad account I just scale good performers there, no matter if it's labeled as testing It's just called "testing" because all new concepts are launched there so they get a dedicated budget and the opportunity to succeed

@max_rosewater I mean, this isn't the same house, lol. but perfectly illustrates the issue with generative AI and LLMs.

Same thing for making 3D mockups of houses that haven't even been landscaped yet. And next, he asked "can we put leaves on the trees." Yes, we can...

AI is coming for every industry... My dad works in the building industry, has for 40+ years. I helped him take his mock floor-plans from a black-and-white draft to a full render. This is just the beginning of what's possible here...