This Week in PHP Internals with Len Woodward

And thanks for making time for me.

Yeah, I can basically just credit Brent Roos for that, uh, because I was watching one of his YouTube videos like maybe 4 months ago and he was talking about, um, different things that the community has. And he was saying, I don't know why anyone's not doing this specifically. And in my head when I listened to it, it's like, I'm pretty well set up to do this. Like I've got the studio, I've got the backdrop already, I've got the camera and the microphone, I've got the lighting, I've got a teleprompter. Like literally the only thing stopping me from being that person, that hypothetical person that Brent is talking about is hitting the record button. And so one of the weeks I just decided to give Claude Cowork access to my email inbox to try to pull down everything from the list and see how it would do summarizing the week's happenings. And it seemed to do pretty good. I mean, that's one of the things that AI's actually good at is summarizing. Um, you wouldn't want to have it write anything for you unmonitored because it's not going to get your, your personality right. Or, I mean, there's not a whole lot of personality in a news digest, but still. But that's, that's how the whole thing started. I, is I basically just wanted to do something that Brent Rous said the community needed.

Yeah. So it's basically like, it's a weekly podcast or news digest. And so we'll go through all of the week's messages on the mailing list. Sometimes we'll bring up something that happened in the internals Discord, but that's not actually an official channel. So we, we try to limit that to just what's on the mailing list. But the big idea is that We needed some sort of proof of life within the community or the ecosystem, because everyone is like, the language moves so slowly, or we're never going to get generics, or this feature is bad or poorly thought out or whatever. And I think the big idea here is that you just need some way to show everyone in the community how much happens on the internals list, because all of these features are just discussed until everyone is nauseated by discussing these features further. Every feature is stress tested. The implementation is reviewed. The, um, everyone has, has input on it to look at the, the syntax and benchmarking and whether or not it even makes sense to belong in, in C or whether it just belongs in userland. Um, This is, this is really the, the big thing is just to show people that things are always happening.

Um, yeah, I mean, generally we're looking at anywhere between a dozen to 2 to 3 dozen separate threads per week. Um, right now we're not really going to be having many because we're in that, that feature cutoff in the cooldown period before, um, like we've got RC1, RC2 coming out and we've got Betas. And like we're cutting off a big version. And so we're not going to be having as much. So my episodes end up being about 10 minutes recently, whereas I have had them be up to like 20 minutes or even like 21, 22. Mid, like around New Year's, I think like January, February, I think that's when we're going to start having the list be quite a bit longer again.

Yeah. I mean, there's, there's been a lot of discussion on how people use AI within the list, like not only for having AI help write the patches, but, um, people handing off their own discussion to the AI, which I think is what people have really pushed back on instead of just using the AI as like a translation layer. Because we've got a lot of people in the community whose first language is not English and they're probably better at writing C and PHP than they are at speaking English. And so a lot of the time they'll take the response that comes in from the mailing list, They'll translate it so that they can understand it. They'll come up with their, with their response and then they'll translate it back and send it into the mailing list. And a lot of the time the AI will still kind of override their, their phrasing and kind of make arguments for them. And so like we've, one of the, the interesting solutions is if you are not a native English speaker and you're, you're doing most of your writing before translating to use more, man, what's the word, deterministic translation tools. So like go to Google Translate, which I mean is still machine learning based, but it's not an LLM strictly, and then just attach your original message along with the translated message. So there is a lot of discussion there on like what the proper approach is for this, what's acceptable, what isn't. And I think that's probably one of the biggest things that's shown up that people might not necessarily have a lot of visibility into, though I'm trying to make that visibility in a way where names and everything might get mostly anonymized so that no one gets singled out. And I'm still not super great at it, but I mean, I'm hoping that I'm doing a decent enough job at making sure that no one's feeling singled out. Because the PHP community is meant to be very large and embracing and supportive and everything, and supporting ELL or English language learners in particular is definitely an important thing for us.

So yeah, a lot of the feedback that I've gotten that's positive comes directly from internals contributors. So I'll, I'll link everything in the Discord or in the mailing list and I'll say, Hey, if I have any corrections to make, respond in this thread, I'll make them, or just respond in the comments here or on YouTube or whatever. And so we'll have instances where I slip up and I call someone by the wrong name or I join 2 names together and Larry Garfield will point it out to me with a smiley face. Or for the first several weeks, Tim Dusterhouse was reviewing my draft scripts before I would before I would record. And that was hugely useful because he participates more than so many other people in that mailing list. And so he had a lot of extra context that I might have missed or that I framed wrong, incorrectly. And so having a lot of people to support that was really good. The way they receive it is really good. They like seeing themselves in stories and getting tagged. And then the negative feedback I've had is usually from Reddit. It's normally people commenting because I post the entire transcript directly underneath the video. So people will just come in and say like, I'm not going to read this. It's all AI. It's like, yeah, I used AI to help me summarize, but I still went through 3 hours of script revisions. I still recorded it. I still edited. I still had to thread everything for every platform so that we could post. But generally the receipt of it seems positive.

Oh geez. Yeah, follow our YouTube because we, even though we've just got the one weekly show, we are ramping up our ability to crank out content. And so one of those bits of content is going to be at least weekly content on using FFmpeg specifically within PHP, because I think at Laracon last year, we had a talk on how miserable it is working with FFmpeg in PHP. We have developed that solution, so we have a PHP extension that gives you language bindings for FFmpeg, and then we want to be able to educate the audience on how to use it effectively. Showing how you would use certain filters, why you would use them, why you wouldn't use them, how to do it through the command line, how to do it through our extension, how to install the extension. So that's going to be cool. And then a whole bunch of other content on making content where we can show like our teleprompter, our video editor, uh, our system audio tools, um, stream capture, and then like all sorts of other stuff too that we've got just for agency tooling. So. Follow me, follow Ed, Ed Grosvenor. He's usually gone by, um, @maybeedward on all his platforms. And I'm usually @projectgopher on the platforms. So follow us on Twitter, LinkedIn, Instagram, all those things, 'cause we got lots of stuff coming out.

So you know the pain.

Yeah, no worries. I'll always make time for you. Yeah.