文字起こし 🔗
11:04:49 There we go. Okay, great. Recently, Oh my god, then it's displaying it.
11:04:56 Well, wait. Let's stop for a second and just say, so it's in the transcript, Chris is becoming an adult.
11:04:57 Okay, there. Go ahead. Yes.
11:05:02 I'm getting to that. I'm getting to that.
11:05:02 I want to make sure that's captured. Now you can proceed. I know I should make sure that was captured because I started with that.
11:05:08 Seems like a very important piece and if we're gonna turn on transcripts I wanna make sure it's there
11:05:12 For the record.
11:05:14 Yeah, I know. So recently I've started, finding ways of getting transcripts of our, podcast and I just got an email this week from Apple that said, hey, we're just gonna turn on transcripts for all podcasts that said, hey, we're just gonna turn on transcripts for all podcasts that are on Apple podcasts.
11:05:32 Yeah.
11:05:33 Oh, Apple.
11:05:31 Apple. Way to go. Although based on my success with Siri. I'm not too confident.
11:05:37 Right, yeah. Yeah. I mean, I'm not super
11:05:39 It's gonna, hey, Seri, what has my first meeting today is serious like it looks like it's not gonna snow today.
11:05:43 Right, yeah.
11:05:44 What? Come on.
11:05:46 Yeah, no, I'm not super, I'm not super convinced with any of the transcripts that we have generated that wouldn't that weren't a handmade but you know that's AI.
11:05:56 No, so, what we're discussing is that, I'm trying out this, this new thing.
11:06:05 Called dressing like an adult because, I'm, conscious of the fact that my style hasn't changed or evolved very much in about 30 years.
11:06:20 Yeah, how does it feel so far?
11:06:21 And as you was saying that I'm rocking my alien hitchhiking to Mars shirt, so.
11:06:27 Yeah, we amongst good company.
11:06:34 Oh cool.
11:06:40 Huh.
11:06:31 Well, so I signed up. And this is not this is not a an add although maybe it should be and I should ask them if they want to sponsor this because I Because they're sponsors of other things.
11:06:46 I don't think I don't think I knew that they did men's clothing as well as
11:06:50 Yeah. Yeah, they do everything. Apparently they do kids. Apparently they do kids clothing.
11:06:52 That's fascinating because I didn't I thought they only did men's clothing so talk about like marketing being super effective and focused, right?
11:07:02 That's wild.
11:06:56 Yeah. Well. I so when I was listening to so I always hear their ads. I always hear my brother and my brother and me advertising for them.
11:07:22 It's not you, it's me.
11:07:10 It's not really their ads. It's that's the macro a brothers. Goofing and mentioning sits fix every once in a while but in the process of so doing they were talking about how their kids are have have stitch fix like things.
11:07:26 Oh.
11:07:28 So like I guess they do kids clothes which is which is wild too. But that's also kind of cool.
11:07:33 Yeah.
11:07:31 Smart. Yeah, I mean you think about like some of those like things you can only wears 2 or 3 times boom and then you outgrow it.
11:07:38 Yeah.
11:07:38 Wow. I don't know. It does the does the reuse offset? The shipping.
11:07:48 Reuse. The shipping is
11:07:47 Or does it matter? It doesn't matter. It's Not the shipping cost like the net pollution.
11:07:54 Oh, no, probably not.
11:07:57 It might, it probably doesn't. Close don't get thrown away. Kids close just end up going to like a through store or something and It doesn't matter.
11:08:04 So anyway, I started, I started doing this thing and then, you know, signed up and like I had filled out the style profile thing like.
11:08:12 Yeah.
11:08:12 Over a year ago and then just sat on it. Because I was like, well, I'm not gonna do this.
11:08:14 Oh.
11:08:17 And then sort of towards the end of last year, I'm like, well, maybe I should though.
11:08:22 Hmm.
11:08:21 Maybe I should make. The new year as the year when I started expanding my wardrobe to include slightly more than just t-shirts in various jeans.
11:08:30 Cool.
11:08:31 So I've had 2 shipments now and this came in my most recent one that came yesterday.
11:08:35 Have there been like hell yeses that have been in the shipments and then like things where you're like, this isn't me at all.
11:08:43 Yes, yes to both. The first, I expected the first shipment to kind of have a lot of misses because they haven't had time to really set up the profile, haven't filled out all the stupid things like they're still getting, you know, but it did have a really cool pair.
11:09:02 Yes.
11:09:00 Of jeans that I really like that are kind of like nicer than the like gap ones I wear every day.
11:09:09 And, in the last box there was a shirt that was like yeah I would definitely buy this off the shelf like I'm Easy.
11:09:16 Yeah.
11:09:18 And then there is, there's definitely been things where like I've put it on and been like, nope.
11:09:22 Or I put it on and Aaron's like, nope, or I put it on and I've been like, well.
11:09:27 This Looks good on me. But I don't know that I want it.
11:09:33 Yeah, yeah, which is totally fine. That's a perfectly finance or for something.
11:09:38 What, how does it work? You pay monthly fee or how does the thing, how does it shake out?
11:09:44 I'm confused.
11:09:43 You okay you there is a stylist fee. And so that you say you sign up.
11:09:50 And you say, this is how frequently I want a box. And there's a stylus fee that's $20 and that that basically pays for the person who picks the things that go into the box and you fill out a stupid like It's not stupid, it's kind of fun.
11:10:04 Around judgment where it's here.
11:10:05 You fill out a, Yeah. Yeah. You fill out a little, you feel a lot of little like.
11:10:11 You know quiz thing that's like would you wear would you wear this sort of thing or that sort of thing what types of things are you looking for what sizes all that sort of stuff and it builds a sort of basic initial profile for you, which I'm sure is going into an AI.
11:10:27
Load full transcript
But not even AI.
11:10:26 And then sometime between sometime between the time when you sign up and say, I want my stuff to come and, like maybe.
11:10:35 A couple weeks before the actual delivery date. They're like somebody on the other side. I assume it's a human it might not be but I it gives the semblance of humanity.
11:10:47 Picks a bunch of stuff And then. Once, once you're past that first month, you can actually get a preview of what's in the box, there'll be like 10 different things and like here here's stuff and you can pick 5 of these things that'll actually be in the box.
11:11:01 And then so you can say, well, I want to look at this and this and this, but not this from this and this.
11:11:06 And in the meantime, again, once you're signed up and you end after the first delivery, cause you don't really have access to the stuff like.
11:11:13 Before. There's like all sorts of like style quizzes that like come up like every day or every other day that like come up like every day or every every other day that just like come up like every day or every other day that just like swipe right for things that you like and swipe left for things that you don't.
11:11:34 And cross referencing whatever data they're getting from that and whatever they happen to have in stock at the time that's also in your size they're like okay well I'll pick these things off the list and shove them in the box and then you get a box.
11:11:44 And if and the stylist fee is then deducted from the total unless you send everything back.
11:11:52 If you send everything back, then you're paying the stylus fee and that's it.
11:11:54 But if, if you take anything, then Silas fees is credited. And then if you get everything in the box, then there's like a discount.
11:12:03 But you can say you can, you know, try it on and whatever and you could say this was too big.
11:12:09 This wasn't my style. This was too expensive, whatever, and you pick the things that you want to keep and you say these things.
11:12:15 I'm sending back. I had to send back, a cool pair of jeans that came in my last box because they were a little bit too big in the waste.
11:12:24 And they didn't have If they have sizes that you can like exchange for, you can say, you know, return but return as an exchange and they'll send that the thing but they didn't have my size or they didn't have the size smaller.
11:12:35 I would have needed. But, but yeah, that's, kind of how it works.
11:12:38 Somebody should do this with DVDs.
11:12:42 Like.
11:12:41 You know, how cool would be like a DVD in the mail?
11:12:45 Like Netflix?
11:12:46 Exactly. Exactly like Netflix.
11:12:49 But, but, but nobody uses.
11:12:50 Which by the way, my mom was like maybe the last person to still be doing the mail. Like I'm pretty sure they kept it just for Karen.
11:12:57 Oh, I think my parents were definitely on that list too.
11:12:58 Yeah. Yeah, my mom was like, I don't wanna switch to digital. I was like, mom, they're just gonna cut you off at some point.
11:13:07 Yeah.
11:13:08 Okay.
11:13:07 And they did? Is anybody, there must be somebody still doing DVD by mail program now that Netflix isn't.
11:13:14 Hmm. Hmm.
11:13:16 Let's Google it.
11:13:18 Okay.
11:13:18 I mean, I remember I remember. In the arts I found a couple different music trading services.
11:13:32 The one that I really liked just called La La and you would, You could you could trade CDs that you have that you don't want and get CDs from other people that you want that you don't have and it was really cool.
11:13:44 That's cool. Yeah.
11:13:47 And I, like offloaded a whole bunch of stuff that way and onloaded a bunch of stuff that way and then if you didn't like it you just like pass it along so it was a really cool way of like like trying out stuff that maybe you didn't know that you would like.
11:14:01 I discovered like a couple of different bands that I hadn't really liked dug into, that way.
11:14:07 But yeah, those things don't exist anymore.
11:14:07 Hmm. It looks like when I googled it, I mean, there was not like a clear answer except that people were like, well, Gamefly does DVDs.
11:14:18 Hmm.
11:14:18 I don't know what their selections like, but. Apparently you can send for game fly and get Tvds or get confused and end up with video games that don't work in your DVD player.
11:14:26 Yeah.
11:14:26 Seems like a win-win to me. Also, DVD players, you know when you look in the tray, which I know you're going to go back in the memory banks because when's the last time you opened a DVD tray?
11:14:34 I, well, I mean, DVD trace the same as it's Blue A Tray and I have a Blu-ray.
11:14:39 When was last time you opened it.
11:14:39 So a couple. Recently within the last 2 months
11:14:47 I mean, I
11:14:45 Who are you? Like this new cloak, like. Are you wearing slippers?
11:14:46 I know, right? I'm old. I look, I am an adult. Yeah.
11:14:51 I have a VCR, so like, No.
11:14:54 Yeah. Yeah, we don't have a VCR anymore, sadly.
11:14:58 We did VCR for a very long time when we got married. Like that was our DVR because we were gonna pay for DVR through the cable or whatever.
11:15:03 Hmm.
11:15:05 So I program that VCR to record our shows and heaven forbid some news broke somewhere we interrupt our shows would be like run later, we would miss the end.
11:15:11 Okay.
11:15:13 And then I would have to hop to the bit torrent to get it and that felt. You know whatever but
11:15:17 See, that's why I just skipped straight to the bit torrent.
11:15:20 This we have this tape in the first like 10 s of the tape you can see like the beginning intro from like 4 different shows like bang bang bang bang and then whatever we were watching because it just cut off a little on the rewind.
11:15:32 Thank you.
11:15:32 It's amazing. We can set like, oh, we watch all of lost by N.
11:15:36 You know, I mean, it works pretty much like a DVR. You could pause and go make popcorn or you know whatever.
11:15:45 That was it was
functional enough and I I think we only got rid of it when we moved somewhere. Move to Jacksonville, so.
11:15:49 I don't know. We just did broadcast for a while and then I think we both had a job and we're like, we're gonna pay for cable.
11:15:58 And, and it, oh no, it wasn't cable, it was dish. And I had a discount with dish because of my job.
11:16:02 Hmm.
11:16:05 And so. I did that and got had DVR built in because they didn't give you an option without DVR at that point.
11:16:12 And
11:16:13 And then you joined the masses.
11:16:15 I only did DVR when I had sling. And it was only, it was pretty much only for like, I only ever used sling.
11:16:23 For soccer.
11:16:28 News Yeah, it's new sling. Yeah. Yeah.
11:16:25 Now, is this the old thing or the new sling? You're the old sling. I guess it's called sling box, but it's on top and it had like little IR repeater and it And it basically was like a Aftermarket DVR and or something?
11:16:38 No, is the app sling. Or the websites thing. Yeah.
11:16:40 Okay, yeah. Well, talking about AI this morning, I have a question.
11:16:49 Sure.
11:16:49 How much data do you need? Do you need to establish a reasonable LLM?
11:16:55 A lot. Yeah.
11:16:55 I ask. A lot. I asked because I have access to an email box at NASA that is open to the general public.
11:17:03 And the comments we get in their art. Absolutely wild. I mean, just. Buckle up.
11:17:20 I mean, I think Yeah.
11:17:13 The, John Hugh public is I'm concerned. I'm really concerned. People asking about hotel accommodations people saying I saw a light moving across the sky in I don't know what country it was, somewhere in Africa.
11:17:30 But it made no sound was this yours?
11:17:34 What? It's what? And it's it has nothing to do with general questions like that.
11:17:41 It's like a Feedback specific to like a piece of the website. It's not like we're asked answering questions it
11:17:50 Wow, it's amazing.
11:17:45 Yeah. Yeah. I, I think that the real answer to your question is that I think that most sort of modern ish LLMs that would do a particular But 2 designed to do a specific thing like I have a big email inbox that I want you to sort through to retrieve things that actually matter.
11:18:23 Hmm.
11:18:17 Are going to be based on existing models that already exist so that has like a base a base set of information like it's already trained on the internet's already trained on like the base level.
11:18:30 And then you add your thing on top and then you say You are now email checker bot and you go through this data set right here and tell me what's important or whatever.
11:18:43 I don't know what's important. I wanted to put together wacky combinations of things.
11:18:43 So like it's a lot. Well, whatever. I mean, whatever, whatever. Whatever thing that you're trying to gather.
11:18:51 Yeah.
11:18:49 There's no use case here to justify and I don't intend to do that. To myself or anyone.
11:18:55 I just want to see what I don't I actually don't think AI could come up with anything stranger than what humans are sending in I guess is what I'm saying.
11:19:03 Hmm, yeah, probably true.
11:19:04 I mean, it's. Wow, it's really something I should I should have a couple of screenshots to share with you because It's top UTF.
11:19:13 Yeah, I mean, we never get, we never get anybody writing into the podcast. So maybe we can just use.
11:19:21 Things that have been written into NASA as a proxy. For questions that we need to answer on the podcast.
11:19:25 I actually wonder if I'm allowed to share these in hindsight. So we may not do that in the podcast.
11:19:31 And what pretend I didn't do it in Slack?
11:19:32 Yeah, I was gonna say I was like, maybe check some permissions.
11:19:29 Yeah. That Yeah.
11:19:35 I, I mean, yeah. It's It's really something and it's like it's a known something like it's a thing people talk about.
11:19:45 They're like, oh, you see this one? There's like back channels on bizarre.
11:19:46 Okay.
11:19:48 Email people have seen commit to NASA. So. There's, there's internal recognition that this is like.
11:19:57 Cuckoo for cocoa puffs.
11:20:10 Oh, there is.
11:20:02 Is there? But like presumably there is a reason that that inbox exists and like like a real a real purpose that it needs to serve like does it ever get anything Legitimate.
11:20:14 Yes. Yeah, it gets real actionable. Yeah, it gets real actionable stuff.
11:20:19 Okay, but there's no way to filter out the the wild and wacky stuff from the legitimate stuff.
11:20:25 Without manual intervention.
11:20:29 Yeah.
11:20:25 I mean, I'm sure that if someone were to put the time in there would be a way to do it, but it's also because it's monitored by like an bunch of people.
11:20:34 Hmm.
11:20:34 It's not my problem. Not like everybody, it's not my problem. It's like it's, you know, like it's a, you know, not my problem.
11:20:39 Yeah.
11:20:42 Also, I think some people would be sad if that Sort of of entertainment went away. Because while we
11:20:54 Hmm.
11:20:52 And completely other news. Totally unrelated. I got this like, 60 MB JSON file of data about like internal websites and Somebody had written the script that crawled all these sites and came up with all this data Some of it seems useless.
11:21:13 And if it's like. Domain. Fully qualified domain. What, I'm not sure what the difference is yet, but anyway, it's a huge file.
11:21:23 There's about 100 data points on each of these things. 40 MB. And I'm like.
11:21:27 But, but like this happened once upon a time. This is not not going a thing. Someone did this.
11:21:32 We have this JSON file. And now it's like, well, what do we do with this?
11:21:34 How do we take this report, turn into something? I'm looking through this, I'm like, I don't.
11:21:37 I don't really wanna like figure out what all these keys are. Because there's like a hundred keys here And to figure out those are, I'm gonna have to read like multiple responses to see what kind of data
is this.
11:21:49 So, I'm not doing that because I'm lazy. I'm a programmer.
11:21:53 So instead what I did is I wrote. I've been doing it in layer of the app.
11:22:00 So there's a model and a model has property. So this property presumably has 100 twenty-ish or whatever.
11:22:03 And what I did is I took that JSON file, pulled it and said, all right, here's all the properties.
11:22:09 Greet them dynamically and now create the migration dynamically. So I actually don't know what the model looks like.
11:22:13 But Blairville does, and the database does. And I started the import before we started this call and so I'm gonna drop it into like Basically like turn it into like a spreadsheet and sort by frequency and say Oh yeah, okay, these pieces are definitely like other models, you know, like.
11:22:31 You know center or whatever and these other pieces are you know strings and these are duplicative and I never need all of these or these are empty and You know, that'll give me some guidance instead of having to figure out these 120.
11:22:42 I can start with. Some of the data but Goodness and not even formatted. Here's a 40 MB JSON file.
11:22:53 Hmm.
11:22:54 Seems about right.
11:22:56 I was gonna say.
11:22:51 No line breaks. Thanks. Yeah. See you open your ID and you're like format that and you're like well I'm gonna get a cup of coffee while that finishes.
11:23:01 I was gonna say you could probably feed it to an AI and you probably could feed it to an AI but 40 40 MB is a lot.
11:23:10 You could probably feed it into like chat GPT and say, tell me. The types of things that are in this without giving me the actual data or doing anything with it and it would probably do a good job with that.
11:23:22 Because it would only look at like the first 100 or so lines and then it would get all the keys and then it'd be like, okay, I'm done.
11:23:27 Yeah.
11:23:27 But like actually telling it to pull stuff from that without I don't know, I've had So I made, I made a GPT.
11:23:38 With chat GPT or like my writing and I feed it whenever I finish a chapter or maybe sometime after I finish chapter, maybe after I finish 2 chapters.
11:23:53 I save a text file of the document and I, Go into the GPT and I edit it and I delete the old text file and I add the new one so it has like a running updated version of the thing and the bigger it gets the harder a time it seems to have at knowing.
11:24:16 Hmm.
11:24:14 What current? Like I like, I, so like I just did it last night. Just mind thinking about it.
11:24:23 Like, so I just finished chapter 11. And so I export the the chat the whole book text file and I say okay this text file and it says like the best way of having it and know which knowledge.
11:24:40 Item to to get into to get the data is by naming. And so I'm like, okay, I just updated are you awake.
11:24:49 ATT to have chapter 11. Can you give me feedback on that? And it's sort of telling me stuff about, well first the first pass it's like very vague.
11:24:57 General stuff that like is about like the concepts that are probably in like the description that I've given it about like what it is like the the sort of prompt that it the pre prompt that it sort of goes through before it even like has a conversation.
11:25:14 I'm like, okay, that's not helpful. That's super vague. Can you give me specific examples from the specific chapter that I am referencing?
11:25:21 And it started going back like 2 or 3 chapters and saying, oh, this happened and this showed this thing.
11:25:27 I'm like, that didn't happen. This chapter. This chapter starts with this line and continues from there.
11:25:35 Can you, can you give me feedback about that? And like, oh, in that case. Sort of.
11:25:39 Okay.
11:25:40 Can I do think like can you say like can you make sure I didn't make any continuity mistakes or are there any weird
11:25:45 I've, I mean, that is one of the things that I'm asking it to do.
11:25:51 Yeah.
11:25:51 I don't know that I am convinced at its ability to actually do that because I think that at a certain point like at this point with like I don't know, 60,000 words or whatever we're at.
11:26:04 Yeah.
11:26:06 It's way too many tokens for it to really have the entire I mean it's it's actually not it should it's supposed to be able to process a hundred 1,000 tokens I think but it's, it's just too much data for it to store entirely in memory.
11:26:21 So I don't know that it's able to to like respond to like the progression of the whole.
11:26:26 What if? Does it does it this is a dumb question does it work better if you say take your time?
11:26:33 No.
11:26:35 I know I do.
11:26:37 Yeah.
11:26:34 I feel like that would be a feature I would want. Yeah. If someone says take your time, I'm like, oh.
11:26:41 This is gonna be fire when I'm done with it.
11:26:41 Hey.
11:26:41 You can tell it. You can tell it to think step by step. And that seems to be like a secret, a magical secret keyword in in doing anything with AI.
11:26:52 It's like a magical keyword that allows it to like, so, so the way that AI kind of sort of works is, is it's thinking.
11:27:02 And so when you do think step by step, it's a, it allows the AI to sort of break down whatever it's doing into steps and then At the end of that sort of reflect on what it just went through and then make a conclusion.
11:27:20 Yeah. Yeah.
11:27:18 I don't care for that at all. That seems too much like a human. Yeah. Okay.
11:27:22 Well, yeah. But like otherwise it'll just like just. I mean, much like me, just vomit words out of its mouth without even thinking.
11:27:31 Except that
11:27:31 Boy, we really missed that. We really missed that as a title for the show.
11:27:37 Without thinking.
11:27:34 Vomit words out of my mouth. Without even thinking.
11:27:37 Without even thinking. Yeah. Yeah, that tracks.
11:27:43 Actually joking aside that is that is pretty pretty interesting that it it speaks to like the nature of of knowledge and like applying knowledge to something isn't breaking it into smaller discrete units you get.
11:27:56 More accuracy.
11:28:00 That's why I use letters to build words.
11:27:58 Yeah, well, and the other thing that Not yet. The other thing that not only I but like other people who actually study these things, have, done and experimented with and talked about is If you, so you ask us to do a thing and it gives you an answer.
11:28:19 Like a coding problem, right? You say, I need this type of thing written in this programming language.
11:28:27 And then you take that output and you say, and you give it back to it and say, okay, can you look at this thing that you just gave me or look at this whatever and make any improvements that you see, that it might need.
11:28:41 And then you feed it back into itself several times. And then at the end, you get something that actually is like, you know, possibly more functional or like, because again, the first thing is like a first draft like off the cuff like.
11:28:53 Yeah.
11:28:52 Not really thinking about what I'm doing and then you but you do you force it to sort of re iterate over itself.
11:28:58 So I, I don't, I do not use chat GPT so because it won't accept.
11:29:03 Kind of do. You kinda, if you're using co-pilot, you're kinda using it.
11:29:05 I don't. Is it? Is it open AI powering?
11:29:12 Well, they don't have my phone number.
11:29:12 Because, because Microsoft, invested heavily in open AI and as a result, it's not exactly Chat GPT but it's like a fork of Open AIs model
11:29:27 It's really fascinating to me. The things that it can do. So my first password I had is, was I had this JSON file and I started copying like the keys that I wanted and I figured well.
11:29:39 You know, it's always suggesting auto complete. So I'll just start auto completing and seeing, you know, if it can pick up on what this file is in the keys and go and.
11:29:49 Oh yeah, for sure.
11:29:48 And no, it's just started making stuff up. Yeah, and it just started adding weird suffixes like Like domain underscore canonical.
11:29:59 Yeah.
11:29:57 The manuscript canonical underscored full underscore domain underscore canonical like Hi, I appreciate this attempt.
11:30:04 But I'll just do it myself.
11:30:05 What I find really interesting, what I find really interesting is I'll ask. Chat GPT to do a thing to write because I was actually doing this the other day when I was goofing off and writing a script to scrape all of Justin Sternberg's gifts.
11:30:22 I would, I asked. Chat GBT to write a thing and it did Actually, this wasn't even related to that, it's really something I was actually legitimately working on for work.
11:30:35 It said like, you know, I was trying to convert a get SSH. You know, they get at GitHub, whatever, to an HTTPS, URL, to an HTTPS, URL.
11:30:53 And I, told chat GPT to do that. And then as I was going, copying that like typing it out into VS code, VS code copilot auto suggested a different thing to do the same
thing.
11:31:02 So then I took that thing that co-pilot suggested and fed it back into chat GPT and I said.
11:31:08 Hey, what's the difference between the thing that you told me and this thing? And it was like, well, My thing is more specific to get hub because it specifically does this thing and that thing whatever and that is more general for like get URLs broadly but also my thing adds dot get at the end and the other thing didn't and you might actually need that if you're trying to clone something
11:31:33 down. So like I thought that was really interesting because like I know that underneath the hood like the models informing both things are basically kind of the same thing or analogous to each other.
11:31:45 Yeah.
11:31:46 So it was interesting to see like, well, one thing over here says this and one thing over here says that and like.
11:31:51 Like when 2 AIs are having a fist fight like who's Who's right?
11:31:56 Yeah.
11:32:00 I mean for
11:32:00 And the answer is the thing that has more context. In this case, that thing had less context, I guess.
11:32:06 I find if I am looking for like a 3 to 5 liner for like, you know. Code that I'm repeating, you know, pretty regularly.
11:32:17 Co-pilots gonna get that and it saves me the keystrokes That's cool.
11:32:18 Yeah.
11:32:21 Otherwise, like, you know, it's
11:32:25 I've been using
11:32:25 It's funny, it's like it's it's The anticipation is great and it works, but if it gives me like a big block of code, I'm like.
11:32:31 I don't trust you. I'm not reading that. So.
11:32:33 So we had we had a dude so we have we've got co pilot fully integrated into our enterprise GitHub and we had a dude from GitHub.
11:32:41 Talk about. Co-pilot and like You know, and I, you know, Kilpilot as a concept isn't new to me, but I've never used previously the copilot chat.
11:32:51 Yeah.
11:32:51 And that was, and they were talking a lot about that. So I started, I started using that after.
11:32:58 To see because the theory is that if co-pilot chat has access to the repository or at least has access to the files so they have open in my editor, then it should give me the most accurate thing because it's got the most context about what I'm actually looking at and then I can actually like insert code from the chat into into what I'm looking at, which is cool when it works.
11:33:23 But I've I've been trying to get co-pilot chat to give me something.
11:33:29 At all worthwhile and it has on various occasions told me that it doesn't have access to the file that I've.
11:33:36 Literally looking at and it has access to. There's a way to like you can do at workspace and if you do at work space it has theoretically access to the entire workspace, all the tabs that you have open, right?
11:33:48 Hmm.
11:33:48 Workspace and I say I've got these things open I'm doing this thing tell me about how this works or how to do this thing whatever and it's like I'm sorry you're gonna need to give me that file for me to know
how to do it.
11:34:01 I'm like, dude, you're looking at it right here. And, then I would copy the whole thing into the chat and like, yeah, I don't know.
11:34:12 And then the other thing that chat can do, which in theory would be great is you can right click.
11:34:18 And you can ask about like a specific line. And or like have it do a thing with a particular line and it will like show you like the code that it wants to insert and give you an option to accept that code, which is also.
11:34:37 Kind of cool. The problem is every time I've done that, It is. Horribly wrong and inaccurate.
11:34:51 Oh, that's pretty cool. Yeah.
11:34:46 I just realized that I can highlight code and tell it to document that code. So I'm just gonna highlight all my code until to document and.
11:34:56 You can also have it. Recently, I've noticed that you can have it right.
11:35:08 But where's the fun in that? That's like the pilot of my day.
11:35:03 Get commit messages. Which If you have if you have integrated Right, yeah, I had it.
11:35:14 I had it do that a couple of times and I'm like, yeah, no, I'm not keeping that.
11:35:17 Gary saves the day by Restoring order to the force by reverting the previous commit that broke everything like I mean like that like I Commit messages are a story.
11:35:29 That's my creative writing.
11:35:41 Well.
11:35:24 I have. I have I have I have team mates that are like. Like there's a certain point after like I've made 60 commits and it's all like trying this thing now like the certain point like I just have given up on like making like legitimately like informative get commit messages because everything is just like well trying this thing nope that didn't work how about this nope Fixing a
11:35:51 typo.
11:35:50 Can I? So can I ask this or this leads me to a question? Because you're doing a lot of CI CD stuff.
11:35:58 And I'm doing a fair bit right now, not a lot, but. If I'm working on something in an action, I'm like, well, how am I gonna test this thing?
11:36:05 I have to push it. How do I avoid like It just feels like testing on prod, right?
11:36:12 When I need to see if the GitHub action works, is there like a good way to, and I could even speak more broadly in CIC, the GitHub actions, it could be.
11:36:18 I don't care, whatever. But Same whole super circle and
11:36:19 I mean, the Yeah. It is a little bit of that. I mean, obviously you're on a branch.
11:36:29 You're not actually pushing to prod unless you are in which case don't. But the way that you really
11:36:31 Right. No, no, no, no, but I, but you know, like, like I said, in a sense, it's pushing a problem because I'm asking, like, run this action now, but there's no testing.
11:36:40 You know?
11:36:40 So, It depends on what you're doing because a lot of the stuff that I do or like to do.
11:36:47 Is essentially telling GitHub actions to run a script or run like various bash commands.
11:36:56 Hmm, test it locally.
11:36:54 And if you can do that, then you can break. Those bash command, you can break those bash commands into a script and run that locally and see what happens.
11:37:05 Or you could have a secondary CI that tests. The script and make sure the script does the thing that you want it to do before you test the thing that the script is doing.
11:37:15 Yeah, it does.
11:37:15 That makes it like before it put into the actions.
11:37:17 But.
11:37:21 Well, here's a here's a good example.
11:37:22 And that's honestly like that's one of the things we're bats comes in help handy like if you have bats tests then you can have bats test and you can have it execute the stuff.
11:37:31 So here's a good example of that. So like, I have to have a test suite that I've written with it.
11:37:36 Well, I need to stand up like You know, an environment to run, I need to compose or install.
11:37:43 I need a database. And right there is when I'm like, well, Now I'm gonna commit 2,700 times today because I have to get the environment file right.
11:37:50 Oh yeah, I forgot the Mysql has this weird blah blah blah. You know.
11:38:02 Yeah, I mean, like.
11:37:57 Etc, etc. And then. Finally, the tests run. Happy and mad that I spent an hour committing to get tests to run like.
11:38:06 Why is it?
11:38:07 And that definitely is a thing. But you could potentially get around that by doing all that stuff locally if you can set up your local environment in a way that sort of mocks what the GitHub environment would be like because I mean the GitHub environment is just the same thing, spin up a box and then install some shit on it and then run through the same steps that you'd be doing yourself.
11:38:29 Yeah. Yeah.
11:38:36 And that, that technical deep dive.
11:38:40 That was fun. Thank you.
11:38:42 I have a kind of topic that we can talk about. For for the last bit of this which is
11:38:51 Reason and you all know this but that our listeners do not I've gotten an email relatively recently.
11:39:00 I kind of ignored it for, or sat on it for quite a while and I continue to sit on it.
11:39:01 Oh, I meant to reply to this. We should have this conversation right now in public.
11:39:03 But but. Yes. Just do it.
11:39:07 But without mentioning them because
11:39:12 Okay.
11:39:09 Yeah, no, I've gotten I've gotten, emailed, from some company that makes I guess a weird drink health drink thing I guess or something and they were like, we love your podcast.
11:39:26 We want to sponsor you. And sponsorship would be good. We have actual legitimate costs for the for running the show.
11:39:33 It's not a whole lot. It just you know making sure that we can pay for the servers and whatever, which is mostly out of pocket.
11:39:40 And the domain and that sort of stuff. Huh, yeah, sure.
11:39:40 Speaking of which, before I forget, can we schedule something? I really want to kill that server.
11:39:43 All you need to do is schedule it. You can keep asking. Yeah.
11:39:46 I know, I know, it's on me. It's been on me for a year when I'd
approached this last time.
11:39:56 Yeah.
11:39:52 Any who? Yes, there's that. But we have actual cost and sponsorship would be helpful, even just a tiny, tiny sponsorship.
11:39:59 And this is the first time that we've gotten offered something really where they're like, hey, we actually want to do this and have gotten, you know, been been in contact and like pinged me regularly and whatever.
11:40:12 And so I kind of said, well, let me take it to the team. And there's a little bit of a negotiation back and forth, but basically they're willing to send us some stuff.
11:40:19 Cool.
11:40:20 If we want and then we can try it and decide if we like it and if we do then they will sponsor us and they will send us money.
11:40:27 We will take the money and we'll probably use it for stuff. And if we don't like it, we can say thanks.
11:40:33 Goodbye. The the only real question is it's kind of a we it's not a thing that we probably would have chosen out of a list of sponsors and I never have looked at sponsorship.
11:40:45 Before this really or thought about it in any sort of serious way beyond It would be cool to have money.
11:40:54 For things.
11:40:56 Yeah, that's the best that's a sentence that applies to most of my life. Like I'm like, oh, people have money for things.
11:41:02 Yeah. Yeah. So we have this offer and it's sort of on us.
11:41:10 I, oh, hold on, let me look at my history.
11:41:13 We have this offer. It's sort of on us to like say, yes, should they send us things or no we're not really interested and I guess that is the question for the floor.
11:41:25 I mean, I, Well, let's talk it.
11:41:29 The other the other sort of side. Tangent to this that I've been thinking about since talking with Allison in Slack over this is is that It's okay to say.
11:41:41 Conceptually sponsorship is cool but maybe not this thing perhaps we could dig into that a little bit more at a later date or now or whatever and figure out like what it would actually mean to get sponsors that we don't that aren't like just sketchy people contacting or maybe not sketchy sketchy is a It assumption.
11:42:02 They're not sketchy. I don't think but just people who just randomly email you over the I know, I know.
11:41:06 This it really reflects on what we think of what we're doing because we're like why would someone contact us?
11:42:13 Hey.
11:42:12 Right, yeah, I mean that is sort of the thing I don't know that I mean I haven't looked at our numbers recently but I don't know that they've grown significantly over the last year or 2.
11:42:22 Wouldn't that be the surprise of it all? We're just like, oh man.
11:42:24 Right. We're secretly like wildly successful. But nobody like our our website is binary jazz.
11:42:27 Yeah. But no one tells us.
11:42:35 Let's listen. Yeah.
11:42:33 Dot com if you listen to us just email us just say hi and that's fine just say hi and and and we'll So Chris sent you, I guess.
11:42:43 I'm pretty neutral. I like I just don't I feel like I feel like I've got so much going on that I don't have a horse in this race.
11:42:49 Sort of. Yeah.
11:42:56 I. I'm against it. I'm only against it because it just seems like capitalism and that's the last thing I want to figure out.
11:43:05 Like.
11:43:06 Because of where it's coming from because of it wasn't something that we chose for ourselves or because taking money to advertise for a thing anything even the thing that we actually genuinely like like for instance if we got a sponsorship from stitch fix.
11:43:27 Is still playing into the capitalist system.
11:43:31 I think it well, let's. I'm not gonna directly answer that because I don't do that.
11:43:35 Yeah.
11:43:36 So let's pretend someone was saying the show was like, I'm going to send you money.
11:43:40 Well, yeah, of course. Right? Because there's no like. There's not a transaction there whereas like, Hey, we want you to try this thing and review it and here's some money feels like there's a obligation to review it in a positive light because we have taken your money.
11:43:58 Well, it wouldn't be reviewing it. It would be advertising for it, which is even, which is different than reviewing.
11:44:02 Yes. Well.
11:44:06 We, whether we, whether we like, I mean, I mean, presumably we wouldn't take it if we didn't like it, but like it's like.
11:44:14 Yeah, it's it's like we're Yeah, advertising. So actually what it is.
11:44:19 Yes, well.
11:44:25 See, I think I think the thing is that like Inadvertently, we advertise for a lot of things.
11:44:33 Yes.
11:44:31 Because we talk about them like I'm not going to go seeking a WordPress sponsorship or WordPress.
11:44:37 Dot com sponsorship. For a variety of reasons. But like we talk about WordPress a lot, right?
11:44:48 Right.
11:44:58 So.
11:44:43 Like that's because that's what we do. So like we're kind of already advertising just because of our own biases and things that we do like, like we've talked about Zeni and we've talked about all sorts of things and like those are advertisements like they're not There we don't have like a segment of the where we break out and say, hey, like that's why I really like
11:45:06 Which is an advertisement.
11:45:02 and I'll mention my favorite podcast again. That's why I really like my brother and my brother and me because Right, which because they literally break out a section they say it's like we're going to go to the money zone they call it the money zone when they do their added experience because they know that like it's stupid and then just for the money that they but they are able to bring in but like they also
11:45:21 Hmm.
11:45:21 have to do it so like it's a whole it's a whole stick. And and and being self-aware in that
11:45:28 I also, well, and I, this is, this is Gary's. Money challenges. I also feel like Oh god, I have to do like accounting.
11:45:36 Hmm.
11:45:36 And that I'm almost like, yeah, go for it. And I could shut up during the money grab or whatever it's called.
11:45:44
Thank you.
11:45:49 Okay.
11:45:44 Because that would be easier. Like as I sit down and do my taxes and go I need to figure out what my expenses were on this server last year like
11:46:03 I mean, I.
11:46:07 Hold on.
11:45:54 It would be it would be hilarious if we did have an advertisement section. That makes me want to do it.
11:46:06 Hi, I'd rather honestly I feel like my general vibe is Okay. It's just our new podcast.
11:46:14 Okay.
11:46:15 I googled this morning. Let me find it. Held around the unix time stamp to the nearest hour and how to survive late stage capitalism, how to defeat late stage capitalism.
11:46:26 Okay.
11:46:30 So this might not be a fair day to asking this question. I'm just being honest with you.
11:46:31 Yeah, I would rather just, I would rather just like auction off terror readings or something and then donate that money to binary.
11:46:41 Or can we just be like?
11:46:43 Yeah. Yeah.
11:46:40 There we go. So you were to hear first if you want to, if you want a terror reading from Alison, just email us, I guess.
11:46:50 If you got, if you got this far.
11:46:51 Yeah.
11:46:49 I even want to go like one run lower that and be like if you're still listening here's how much we spend on What's domain?
11:46:57 $9 a year, $12 a year. Well, there's 2, cause there's dot us and.com.
11:46:59 So.
11:47:01 Let's call it, let's call it $25 a year. Let's call hosting.
11:46:56 Yeah, 9 or 10 or something I, right. Yeah, we are doing. Sure. Hosting probably 20 bucks or something.
11:47:10 Well, I think it's, I think.
11:47:09 Oh, maybe not. You've, you've got the, you've got the CDN too.
11:47:12 It's, it's on a drop by itself. So last year it was. 60 bucks, but this year, whatever it moves, it'll be.
11:47:19 I mean, it won't matter. It will be insignificant. Oh, there's an S 3, so maybe, I don't know.
11:47:23 Yeah, yeah.
11:47:24 It's not a lot. But we should just put that like. We need, well, okay, I'm a developer.
11:47:31 Of course we're not gonna do this easy way. We need to build something that hits the API and gets the expenses.
11:47:40 Hmm.
11:47:36 From these services and post it live on the site no no this is how this is how we're gonna deal with this so we could put in the corner You know our current year today cost is this So far, people have offset it with this.
11:47:48 If you'd like to offset it. You know, here's how you can do that.
11:47:51 I mean, here's the other, here's the other real thing is. I think sometime last year I opened up our GitHub.
11:48:01 Profile, GitHub. Com slash binary jazz to donations. We've, accumulated exactly $0 so far.
11:48:09 But it's possible.
11:48:12 Do you back any projects on GitHub personally?
11:48:10 But it's there. And you can sponsor. Hmm?
11:48:17 I personally do not back things on GitHub. I probably should. But I
11:48:22 I hate the recurring thing, but I do I have like a list I go through it from time to time through throw coins into coffers.
11:48:29 That's nice of you.
11:48:34 Yeah.
11:48:31 It's the least I can do given like without those projects I would not be eating. So.
11:48:38 I am, where was it? There was something, some task I was doing and I came across someone's solution for it and it was another freelancer and he had like, like buy me a coffee or whatever and I sent him some money.
11:48:53 And cause I was just like honestly it saved me maybe 10 h of time to find this code snippet.
11:48:59 Hmm.
11:49:00 Yeah.
11:49:06 Right, yeah.
11:48:59 I was so delighted. And I was just like, oh, I'm definitely sending you money because it was, it just felt like hiring a consultant or something like just bringing, bringing someone in.
11:49:10 And he was so delighted and I got so delighted by his delightment about it. Like, no, it's nice.
11:49:16 I've gotten donations before for various things that I've done because I think in my like the reed.
11:49:26 Dot txt and some of my plugins it like has a space for like donation link or something and I set up like a PayPal.
11:49:31 Hmm.
11:49:31 Dot me like URL or something and I've gotten
11:49:33 Do you remember in the.org repo with plugins that you also they would ask you like where you can how you could provide like a support link like that.
11:49:42 Yeah.
11:49:46 Hmm.
11:49:43 Would always like suggest like find your local I don't know, animal shelter or what was I big into at the time?
11:49:55 Hmm.
11:49:53 Oh, foster care. That was important to me at the time, so. That was that was what I suggested.
11:49:58 Find something local. Give that money. Tell me you did it. That way you made me happy.
11:50:03 Yeah. No, never did. So, probably just more about my plugins than anything else. My plugins were actively detrimental to foster children.
11:50:13 No, no.
11:50:13 That's Where I was. Yeah, like I told you, this is the DVAF mode today.
11:50:20 Okay.
11:50:24 And on that note. Where you can say, Gary. So you open your, opening your mouth.
11:50:31 Now it's closed.