Is Testing Dead? and The Younger Generation (of open source maintainers) These Days!

No talking about Things until...

'Til Until the allotted time.

Till the allotted time. Because talking about things prior to the allotted time is

It's a bit of a waste, really,

isn't it? Yeah. Yeah. When you think

Unrecorded talking about things.

Unrecorded talking just didn't happen.

Yeah. Exactly. If it's not on the podcast, it didn't happen. Well, that that's not true either because there's a lot of life that does happen that is not on this podcast.

Yeah. People say if it's not in JIRA, it didn't happen.

Yeah. If it's not on Strava, it didn't happen for for the runners out there.

That's true. Yes. Except for the, marines at the dark side of wherever it was where it was on Strava, and it made a nice highlighted outline of the of the secret base.

Yeah. It's it's an interesting privacy and security nest of vipers, isn't it, Strava? Because that's actually, it's it's got quite a lot

of your habits in there. Someone might call it sensitive personal information Yeah. That you publish.

Yeah. So I would imagine in the terms and conditions for using Strava, it's buried in there somewhere. That's it. Your running data is their data to be used as they see fit.

Well, they are. They are a a Silicon Valley corporation,

aren't they? No. That's true, actually.

Are they? I mean, actually, are they a Silicon Valley?

The Silicon Valley, but the social media just with a particular twist.

Strava is based in, drum roll, please, San Francisco, California. Alright.

Okay. Yeah.

Yeah. It operates as a global company.

It's a global company.

They're all global companies. Yeah.

Well But, you know, I'm happy to trade virtually all of my privacy for someone to give me kudos on the run that I've just done, which is part of the modern condition, isn't it? You know?

Yes. Who was it who said, I'm about to do one of my terrible misquotes, so I'm going to try and get it right this time. Benjamin Franklin's famous liberty safety quote lost its context in the twenty first century. So Benjamin Franklin said, those who'd give up essential liberty to purchase a little temporary safety deserve neither liberty nor safety.

Oh, spicy. So he he wouldn't be particularly happy with, happy in the modern era, would he?

No. No. I I He'd

just be shouting at everyone all the time.

Yes. Maybe he'd work in the technology industry

Well, yeah.

Shells and clouds.

Yeah. That's true, actually. Yeah. Yeah. I'm sure Meta would give him a job to quieten him down a little bit.

But apparently, Benjamin Wittes, a senior fellow at the Brookings Institution

Okay.

And the editor of Lawfare said to NPR's Robert Siegel that it wasn't originally meant to mean what people think. He didn't imagine a future of cell phones. Yeah. I think we should probably get to things.

Things you say?

Things I say. No. Well, things. No. Not just things I say, obviously. Things that either of

us Well, if it was up to you, Ian, I was on mute at the start of this recording. Yeah. So it would be just be things that you say.

I can't believe you fooled me into taking you off mute.

Yeah. I thought when I listened back to the podcast, it had a distinct lack of me in there.

Yeah. Yeah. Yeah. That would be your your feedback when you when you heard the podcast. I'll just do an edit. It's like, what a lot of things without any of Ash's contributions. Yeah.

So on mute, I don't have access to any of the buttons to make silly voices or play. So Ian says that he loves democracy. Actually, I've never heard you say that. But I imagine Ian saying that he loves democracy, but do as I say.

I'm not giving up any of my freedom to give you a little security or something.

Yeah. Freedom for me, but not for you. Peculiar idea of freedom.

No. No. You you you you're free to buy a little box to to to plug them on for

you to that

you can do.

Like that. Freedom freedom, purchase freedom.

Yeah. Yeah. Yeah. You're free to

I'm free to purchase my own freedom.

I've paid for this freedom, I tell you. Oh, dear.

So things Yes. Yes.

I think it's my turn to go first.

I think it is. I think it is as well.

Without even consulting any kind of algorithm. No.

No. I I actually remember it.

Well, that's bizarre, isn't it? I think the last episode came out, what, a week ago yesterday. Wow. Wow.

That's pretty good. Yeah? Back on track. Back on track. Somewhat.

Well, yes. We'll see. Yeah. We'll see. But, yes, this could be could be a short gap episode. Short gap episode. Short episode gap. Okay. So if I'm going first

You are.

I'm going to plow ahead with my thing. Okay. So this thing could have been called a number of titles, but in honor of the fact that it's you that I'm talking to Okay. About it, I called it, is testing dead? So it's not a real question because quite plainly it isn't.

Yeah. But you didn't this is different, isn't it? Because you previously, the statements were testing is dead. They were said as statements, not questions.

Oh, okay.

So you you've not quite met the criteria for me to go and write a blog post refuting that testing is dead because you've asked the question, is testing dead? And I could say, no. It's not.

I could edit the title.

Well, yeah. True. True.

Just a slightly different word order with different punctuation.

But you could have it, testing is dead, with a question mark at the end.

How many ways could we pronounce that?

Yeah. Yeah. So why is testing dead, Ian?

Well, obviously

Why what prompted you to ask this question?

Obviously, it isn't. But the subtitle which I came up with, which more accurately reflects this

Okay.

Thing is AI code based comprehension.

Oh, right. That's more dull. Yeah. Yeah. It is, isn't it?

It's extremely dull by by comparison. But I came across something that has kind of blown my mind a bit.

Okay.

And to your and my and everybody's amazement, it's about AI. Yeah? You weren't expecting that, were you? No. Nobody expects anyway.

So

this was from a blog post by the most excellent, Simon Willison.

Okay.

And I'm going to make the bold suggestion, if you're interested in this stuff, that you using the magic of really simple syndication or RSS Mhmm. As we like to call it, you should subscribe to his blog. And happily, I'll make that easy by putting a link in the show notes because that's the kind of guy that I am. That's the guy. And his blog post was entitled o three mini is really good at writing internal documentation.

Okay.

I'm like, oh, is it? So I had a play. And so he's basically made some open source bits of software that make this very easy from a command line.

Mhmm.

So he's got a command called LLM, which is just a What

does that do?

What does that do? Yeah. So it's almost impossible to guess. So you type l l m, and you might say minus m o three hyphen mini.

Okay.

And then you can pipe data into it and get a response back. So it's like a Unix pipeline command that will basically let you pipe things into and out of language models. And so that immediately means you can write some really interesting kind of scripts or whatever Yeah. Because you can it just it when it turns into something that's just in your pipeline, that's quite interesting. But that's a kind of a bit secondary.

Yep.

And then converts them into basically a a data stream. Not that I'm making it sound more mysterious than it is. It basically just wraps them in some delimiters and fires them out the standard output, which you can then pipe into an LLM.

Okay.

So what what you could do is you can do files to prompt and then make a list of directories in the top of your repository that you've cloned. I often include package JSON and readme dot MD and and s r c if I've got that. But you're just where your source code is.

Okay.

Don't you don't just do the whole thing because then node modules get sent to them. That brings your query to a rapid close because node modules is too big. Even for o three Mini, which I'm quite grateful for because it probably would have cost me a fortune if it hadn't been too big. True. So you basically use files to prompt to kind of slurp up your repository, and then you pipe it into an LLM and you give it a system prompt.

Right.

Hopefully, it doesn't think it I meant it should use markdown for the testing for the actual testing. And lo and behold, out came a markdown document Lo and behold. With a comprehensive test strategy in it. And I know it's comprehensive because it says it in the title.

Yeah. Exactly. So I I didn't read any further.

This was this was my definite attempt to prove that possibly testing was dead. Right. So I look forward to your perspective on the the fate of testing, but then we can maybe talk about the broader thing.

I did look at this test strategy.

And I can tell from your face that you were impressed.

I was impressed. I was impressed. It says a lot of things.

A lot of things.

It does say a lot of things.

Topical.

Yeah. Exactly. And most test strategies I've ever seen say loads of things. Yes. And they list loads of technologies, but they very rarely say what the mission of the testing is.

Well, you wouldn't want to cover anything hard.

No. Because

that might make it bigger, yet bigger.

Yeah. And also feeding a code base in to and asking how to test it. Basically, all it will come up with is how do I automate what already exists. So basically, a test strategies like this tend to wrap applications in very hard shells of automation, which are very then kind of change resistant. So this isn't like if I saw a person had come up with that

Yes.

I see people come up with that all the time, and that's why the LLM came

up with it. Yes.

Yeah. Because it's just what people think testing is, as in it has no mission. It has no guidance on how to select which tests are most appropriate where, which is like tons and tons of the battle. Just using a code base to determine how to test without any other context is very shallow indeed.

Well, there's no there's not no other context. So the ReadMe, for example, has a reasonable explanation of the application in it. Yeah. Just for the record, I don't think I I believe testing is not dead. I did not have high expectations that you would come away from reading this with applause and wonderment. Yeah. But I suppose

I mean, it it looks to me like it's just regurgitate what was in the read me as part of the overview, and then the rest of it is, like, these are all the tools and

You're not telling me that I would have used the word full featured in my overview.

Well, it's done an LLM thing, hasn't it? So It's dealt all things my readme. Into the treasure trove of your readme Yes. And come up with a full featured explanation. So so, yeah, I don't Full featured explanation. So I I have seen lots and lots of test strategies like this. Yes. And I'm not amazed that another line would come up with it because I've seen lots of people write them as well.

But equally, I wonder what it would come up with if we gave it a prompt that instructed it to do the things that you've Yeah. Said. So you have those two questions about what's the mission of the testing Mhmm. And the other question.

But it wouldn't know that from the what you fed it as in the code base. So It would guess. It would guess. I reckon it would guess.

It might be right as well, but it might not be.

Yeah. Exactly. It'll try something because that's generally what LLMs do, isn't it?

They give it a go.

Give it a go.

Like the song? Yeah. But I've never never played you the song, have I?

What? The LLM give it a go song? No. No. But there's a

when I went to What a very specific song. It is. Yeah. I went to Perhaps we should write that. Croppity Festival. Right. Okay. In last August, I'm gonna say Yep. And listened to an Australian men's choir called the Spooky Men's Chorale.

That's a great name.

And they had this marvelous song called We'll Give It a Go, which was all full of things that they would give a go to, like children's party face painting and wind turbine maintenance

Okay.

And air traffic control, to which they always returned to the chorus and said, we'll give it a go. Anyway

Yeah.

Yes. That's what LLMs do. They do

it again. Absolutely. So with with no information, no no skill. No. But they're they're keen.

They're keen.

Yeah. So, you know, that's good. So that was my general thought on this test strategy. Yes. I mean, it's not, like, bad, but it's just, like, mundane.

Well, yeah. And the thing is that it took two seconds. Yeah. Well, actually, that's a lie. It probably took fifteen seconds.

Yeah. But it talks very confidently.

It does. Well, that's what they do.

You know? But Which, you know, is probably good in terms of test strategy because everyone would go into the meeting and look at this, and the LLM would be so confident about it that everyone would be like, yes.

That must be right.

That must be right.

Sounds sounds amazing.

If you can say it with that much confidence, then surely it's correct.

I thought it was an interesting experiment. I only did it because it was you. Yeah. And I also did another thing because I didn't I realized that we would be in danger of Yeah. I

mean, we should caveat here that every test strategy I meet, even probably the great ones, you know, I'm I'm a bit picky about them.

Well, it seems like you have some views about what they should include.

Yeah. Yeah.

So I think it will be interesting to try it again

Yeah.

With that information to see to see what came out. But I yeah.

Yeah. So I think we'd like a couple of useful resources, let's say, for how to how to pick what sorts of tests cover your application well without writing ridiculous amounts of to end test automation, or whatever it is. But that's a skill. But people have, like, enumerated that skill, if you like, and how to make that those types of choices.

It does refer to the test pyramid.

Yeah. That's okay. I'm not against the test pyramid. Just don't do it. Just don't take it too literally. It's like a lot of things in life. Right?

It says we will follow the layered we will follow a layered approach. And then it explains what unit tests, component tests, integration tests,

and end

to end tests are.

Yep.

Simulate real user scenarios.

Yeah. Got some issues with saying real user.

None of your fake users

will be simulated. So

You know? But the other the other thing I did with it on the same code base was something which perhaps it's more suited for, which is I asked it to create a new developer onboarding reference Yep. For the repository. Okay.

Cool.

So that would explain to a developer coming along what they would need to know to be able to contribute to the

Yep.

The code base. And I thought it did a reasonable job of that. Although, when I looked at it, I had small moments of embarrassment about things in the code base. Now that's still there, is it? It is. Kind of moments.

Does it have a watch out for these workarounds section?

It it doesn't, but might. It might.

So it might be well served to have.

But yeah. So for example, it goes through the use of server actions to do the back end forms of validation Mhmm. With Zod and schemas and what have you. So it finds all the things, and it's I think somebody reading that would have a much better chance of understanding the code base than maybe me wittering at them for five minutes about it.

There's probably, like there's a lot of read mes out there on GitHub that it's probably been trained on. Yeah. And there's some good ones and there's some bad ones, you know, but when you look at when you look at this, it definitely has the, like, the marks of a of a good one.

Yeah. I think it's a d it's a decent one. And it's kind of I read down it thinking, oh, yes. It does that. Yeah. Oh, yes. That's in there.

Yeah. Yeah.

So I thought yeah. I so I thought it did a reasonable decent job of that. And I've done a lot of other things with it as well. Yeah. So I've done things like asking it to conduct security code review and to identify functions that maybe were unprotected, that shouldn't be, or Right. Attempted stuff like that. And it's done a really great job at at that, and I've patched a couple of things based on based on feedback from its its review comments.

Yeah.

So all in all, it kind of blows me away that you can now feed a code base to an LLM and get something sensible back. Yeah. And I guess the limit is really how good are you at writing a good prompt for it.

Yeah. Yeah. I think it's like a question of, like, what like, with all things as an LLM, it's a question of intent, isn't it? So, basically, the things that LMS based so generated test strategies always miss is, like, miss things like risk and value. You know, hard the hard questions as in, why should I bother writing a test for this if there's no risk around it or it holds no value?

Well, no. I mean, the thing is that you could treat something like that as a starting point. Yeah. Because it's written about things which are important and not important. You could get rid of the not important things and add in some more important things. Yeah. Because working with a document that's bad and making it good is always easier than just having a white blank piece of paper. Or it's often easier than having Yeah.

Yeah. Sure.

A white sheet of paper, which is almost entirely irrelevant because no paper has been involved. Paper? What's that box paper?

Yeah.

But, yeah. So the other thing that I found has made it a lot better is making good readme dot m d's in the roots of my project. So in this one, I got some very high level user stories to say this is what it you know, here's an overview of what it is. This is what it's supposed to do in the form of these these user stories Yeah. And some stuff about the technologies that are being used in it.

Oh, is it? In

the wake of r one.

The DeepSeek, please look at us again.

Yeah. Yeah. So deep We're over here. DeepSeek r one. Oh, look at this. It's much better than o one. Oh, here's o three. Here's o three, which is actually also much better than o one.

So what makes it better?

My sense of it is it's smarter. And I think it also I'm going out on a limb, but I think it has a much bigger context as well.

Oh, right. Okay.

So you can fit a lot more in. Mhmm. And and it's higher in benchmarks and what have you.

Yeah. So it's

seen as a decent advance, but it's it's just mini. There's no o three full fat.

Yeah.

It's just o three mini

at the moment. So you know the the blog post by Simon Willison? Yes. So and the tools that he's he's written? Yeah. So, literally, you could have them in a in a pipeline because they're just

Python.

They're just, yeah, they're just shell commands you can run out there.

I installed them on my machine using PIPX Mhmm. Which is like a version of PIP for Python packages. Yep. Testing the plosid resistance of my iPhone at the moment. But PIPX basically installs a whole Python environment for each command that you install with it.

Because one of the challenges with any documentation is somebody changes some code somewhere, and then your documentation is then out of date. Yes. So

So you you reckon a a git commit hook Yeah. To recreate the documentation Yeah.

That kinda thing. Commits. Yeah. Might be actually be very, very useful.

Yeah. Yes. Yeah. I think you're I think you're right. Yeah. Especially if you gave it the documentation and told it to update it Yeah. So that it was at least reasonably stable and not inventing it. Yeah.

Because these are the things that teams find really hard. Yeah. It's like writing good documentation. A few years ago, I worked on a team who had a technical writer, and it was awesome.

Yeah. Yeah.

So the runbook and the general documentation around the service was incredible that, you know, not every team has a technical writer.

Well, if you did your documentation in markdown and kept it in the repository, I think you could, you could do quite well with something like this. Yeah. This could be productized in a way that, at the moment, it's just me typing command. Yeah. Yeah. I'm suitably impressed.

So there's other tools, like, older tools called, like, Code Scene, which does analysis on your code base, even just, like, tools that do cyclomatic complexity checks and things like that. So do you think o three minutei used in this way gives you a bit more? So if you said to o three minutei, here's my code base. Please tell me about where there's areas of complexity of this order. Cyclomatic complexity is like how deep do your methods go, isn't it?

So I don't know about that particular one. I mean, the thing is that I my inclination is to not use AI where you could do something algorithmically. Yeah. And the reason for that is that it's one of those things when you start taking climate and emissions into account Mhmm. On these things, it's always never a good idea to use a a LLM for anything that anything else can do Yeah.

Yeah. Right. Okay. Yeah. Because things like looking at code based complexity metrics, they'll tell you where it's complex, but that's where it ends because it's algorithmic, isn't it? It's just like Yeah. Here is complexity, and it's up to you whether or not you deal with it, basically. So I don't know. Would would an LLM be able to give you options about how to deal with it? Well, I suppose it would, wouldn't it?

I've got a report that it wrote about where I had authentic auth authorization code at the front of server actions where I haven't. And and so it went through, looked at them all, and then came back with some points. But basically, it said, based on our review, it's the royal wean, the security of the server actions in this app is generally robust and follows these guidelines, role enforcement, blah blah blah, public versus authenticated endpoints, blah blah blah, middleware enforcement, blah blah blah, logging and error handling. In summary, every server action that requires authorization is checked at the appropriate level. The design that restricts administrative operations to super admin users while allowing regular users or unauthentic so it's it's understood what's going on Yeah.

Yeah.

It's very it saves a lot of very boring

Yeah. That doesn't sound like much fun.

Work.

Going through that particular process sounds a little bit dull.

But, yeah, I I do think that this is the kind of thing where knowing what good looks like everything to do with AI, knowing what good looks like is critical. Because if you don't know what good looks like and you can't evaluate what comes out of it because you're just in a domain you don't understand sufficiently

Yeah. Yeah.

Then you run into horrible risks of it lying to you.

With total confidence.

With no malice, but total confidence.

Yeah.

And that's why I can look at that test strategy thing that came out and say, oh, that looks like a a lot of test strategies I've seen before. And then you say exactly the same thing, but in a much more disapproving voice. Yeah. That's true. But

yeah. So is is testing dead?

No. Of course, it isn't. I never thought it was. Is it just, like, it's just good clickbait, isn't it? Well, what we'll do is we'll include it in the title of this app, and then, like, every tester in the world will listen to it. We'll have unheard of listening figures because people like to listen to something that makes them rage. Oh, dear. I mean, yes. That's a good it's a good trick, actually.

We'll have to see. Just have the most controversial titles.

By the way, testers who listen to this and have arrived at this point, I'm really sorry for the transparent manipulation that brought you here.

And we all know that artificial intelligence will replace developers and not testers, don't we?

Well, I've got a story about that as well. So I'm using a IDE called called Cursor.

I've been fired Yes. And replaced by AI.

I'm using an IDE called Cursor Yeah. Which is an, a fork of Versus code that's had AI stuff put into it. But they've put a kind of agent thing into it now. So you say something like and it was some some big task. I would like to move all the thing into the other thing or something like that.

a genius.

Yeah. I'm a genius. Still a genius. And it listed Contentful Alright. And Sanity and Strapi Yep. Which are three CMSs, content management systems. And then it talks about them for a bit, and then it said, I recommend you use Contentful. And I said, well, I think there'll be a team of three people who will want to update it. And he said, oh, they've only got two on their free tier, so why don't you use Sanity instead? And then I said, okay. Can you just do it for me? And basically, it did.

Right.

It needed a bit of help sometimes.

Mhmm.

And it installed the studio thing into my repository, and then I was able to access the studio thing and and write. And then it created a schema for me for the various gaps in the home page where we want to put some content. And then I put the content into the into the CMS, and it it worked. It actually integrated a CMS really easily.

Right.

And then and then I said, well, all of these feature links that you've put along the top, I want to make a feature route, and then I want you to change the way the the feature thing is structured in the CMS to then use it to be able to if I add another page in the CMS, it will add another

Yeah.

Link. Just did it. And again, needed some help and there were some problem bits particularly because I'm insanely decided to use Tailwind four, which has only just come out. Right. So the the LLM is like, right, so I'm just going to write this into your Tailwind config and then a Tailwind config version Tailwind version three config file would appear with this stuff in it. And I'm like, nope.

It's not

it's not Tailwind three. And then it also it also just like

really gotta get it done. Use the latest version. Use the experimental version in the library. Version. Library. Experimental version.

But the other thing it it also kept doing was I have a next dot config dot t s Mhmm. Because I like to use the type stuff to make sure my config file is right. And it kept writing a next dot config dot j s with its own things in it. So it was stuff like that where it made mistakes. I mean, it it was definitely not perfect, but it became a conversation.

It was adapting.

So, yes, maybe, developers are closer to the hijack Yeah.

Yeah. Absolutely.

Than testers, given the, test strategy failings of AIs.

Did did did did Cursor change the tests as well for the new or is there no tests?

There's no tests. I I what I need to do is say to it, please do tests. The thing is

Just like developers.

I don't think I'm yeah. Exactly. Please do tests. The thing is I don't think what what I've learned about testing is that it's harder than I think Yeah. Even when I take that into account. And, the very small amount of tests we've added to Ilkki Live has been very hard. Yeah. And it's hard for me to be sure we're doing the right thing. So, yeah, testers. Testing is hard.

Well, like I said, it it uses concepts such as risk and value, which are very difficult.

But I'll do to the AI and say, do tests. Do test. Do test.

See what it does. Yeah. Worst prompt ever.

Yeah. Yeah. Do yeah. Test. Test. Maybe I'll just shorten it further. Just test. Test. One imperative syllable.

It would definitely have a go.

It would give it a go. Yeah. Go. Go. You give it a go. Go. Go. It's a very good song. You should definitely listen to it.

An excellent thing. A thing that I enjoyed.

Well, And, I don't feel too beaten for my, advocation of, of AI test strategy. And we've got a really good click baity, title that means that thousands and millions of testers will be unable to resist listening to it.

And and

And just imagine the blog posts that come out.

The latest in a long line of people to declare.

The death of

The death of testing. Yes.

Which to be fair, neither of us is the latest in that long line of people. No. I think we are quite clear between ourselves.

Yeah. I'm sure someone will come out with it in the next

twenty minutes. Is Brian Blassard. Testing's a lie.

He'd say it. No.

It would be better than what I just produced even if I made it into a I could try it with a silly voice.

Testing's alive.

That's just stupid. So stupid. In fact, I'm very likely to edit it out.

Keep it.

Keep it.

Keep it. Should we have a, an interlude?

Yes. Yes. We shall. One of the clagger voices last time was definitely aimed at saying, you're going to get so sick of this piece of music. That's literally one of the clagger voices.

Alright. Okay. It's just too relaxing.

Yeah. We'll wake up tomorrow morning just drifting. Sitting in these chairs going, oh my god. What was this?

It's just nice and warm.

Is it?

Nice and warm in here. Yes. Nice relaxing piece of music. Yes. Been at work all day. So should we talk about interlude things?

Yes. Sorry. I can't resist the scratching noise as well.

Yeah. It's good.

So there are some interludey things. Yeah. We didn't think there were, did we? No. But then we found low. There are.

Low.

Low. That's we we should say that more often.

Yeah. Try and reintroduce it back into the, Lexicon. The common lexicon.

The common lexicon. Mhmm. I think there ought to be a common lexicon. I think that's a that's a good idea. So I'm still clicking away to try and find episode 28 so I can find out what any of them are. Since I first heard of Baldur's Gate three on this podcast Yep. And I announced my playing it, I now can announce I've nearly finished it.

Nearly?

Nearly. In other words, I've not finished it. But I am fighting the final battle, which is very exciting. Yeah. And so far, I haven't done anything too morally repugnant in it.

Alright. Okay.

Which means that I'm now a mind flayer.

Oh, okay.

Because it was either me being a mind flayer or me making someone else be one.

Yeah. It's interesting games that have options to be really, really quite evil. Yes. They're hard to take, though, aren't they sometimes?

Well, it depends. But, yes. Mhmm. You kinda think, if I took all these evil options, it would make the game quite interesting, and it would be interesting. But would I be reducing my brain's resistance to evil options in a way that would spill out into the rest of my life?

Yeah. It's it's kind of consequence free evil options, isn't it, in video games? Yeah. Yeah. Well, apart from usually, I find you if you take the evil option in a video game, it just makes it harder because, generally, you're a bit more alone.

Yes. Yes. There was a

Because if you just continuously turn on your teammates and either kill them, maim them, betray them, however you decide to do it.

That's why you need to be a character with high charisma so that Yeah. After you do all that, they they still stay with you. They don't know why. I'm still here, and I don't know why.

You're such a jerk.

Yes. But what are we doing today? Yes. Yeah. That. So, yes, I have tried to be as good as possible. Mhmm. But it is it is interesting because it presents you with choices. This is, I think, what you were so impressed with when we were talking about it before was that each of these choices has significance, which means that by the time you plot out all of the Yeah. Directions the plot can go in, it's a lot.

Yeah. I think that a lot of games in the past have sort of touted having meaningful, impactful choices. And then

the way

that the game plays out, it doesn't really matter what you do, Which is always a bit of a shame, isn't it? Because it's like it kinda breaks the immersion a little

bit. But yeah. So I'm still loving it, but I am near the end now. Yeah.

So will you miss it?

Well has

gone?

Civ seven has arrived now. Yeah. True. So I'll probably miss it Yeah. A bit, but I'll probably play it again.

It's the I think it's the mark of a good game. When you finish it, you're like, I'm sad that I'm not gonna be at least for a little while, that's not gonna be, you know,

not gonna be with these made up people.

Yeah. Yeah. Absolutely. Absolutely. There are games that you miss, I think. Books.

You know?

Yes. You you finish things, and you're like, oh, I'm sad that I was finished. Yeah. Yeah.

Jonathan Strange and mister Norrell Yeah. Is my favorite book for that. I always regret finishing it and wish it wish it would go on.

Yeah. Get writing.

Get writing. I'm not gonna do it. I paid a very good £12 to be entertained by that that book. It probably was more than that.

But Yeah.

So, yes. That's my my Baldur's Gate three news.

So I'm going to well, Gwen and I are going to Stratford Upon Avon No. On holiday.

No. How can it be?

Yeah. This is one of the the the perks of being with, Australians because they're just, like, you know, places that you probably wouldn't go in The UK. Exciting. Seem very exciting. To them? To them. I'm quite excited as well. Oh, there

you go.

I've not been since, like, a school trip many, many, many years ago. So

I haven't even been on a school trip many, many years ago to that. So

because it was just like, you know, when everyone had to do Shakespeare, which, you know, wasn't a bad thing in at school. But as I think as as I've got older, I've kind of become more interested in it and realized that I actually quite enjoyed it when I did. Hamlet was the chosen one at at school for me. Yes. So yeah. Yeah. So I'm looking forward to going and doing some touristy things. I think also, as I've got older, I've I enjoy touristy things.

I think that's just I I think that's that's just because you get older. Even you think back to your parents and things they were interested in and thought were really great, and then you're like, please don't make me do this. But now you're like, oh, a National Trust property is good. Off we go. True. True. Rip it home.

Yeah. I think we might go to Warwick Castle.

Yeah. That's a marvelous place. A friend of mine got married there, actually, and I went and attended. It was rather Nice. Marvelous.

So, yeah, we've got a few days to explore that general area. I don't often think of the Midlands. Well, because you you have, like

The the flyover Britain. Well The train throughs. Yeah. Yeah. Train through counties.

Exactly. So it's like, you know, you go to Leeds, you go to Manchester sometimes, you go to London, and then you do that triangle quite a lot in, like, your general working life.

You do. You do.

So maybe there are other places that exist that could be seen, and I will find out.

And see them.

And see

them. Let them try not to be seen.

Let them hide. I'll find them and see them. So, yeah, Stratford Upon Avon.

Well, you can tell us all about it I will.

In episode 29.

I will. So last time you were telling us that the atelier is kicking off. How's it going?

The date is the 05/13/2025.

Is it at the dawn of the day? It is,

09:30.

All folk music happens at the dawn of the day in May.

Alright. Okay.

Because it rhymes.

Even if the music doesn't actually happen. Well Even if they have a lion.

Even if it's a bit of a stretch

to

say it's the dawn of the day.

And we've got some speakers. We've got some sponsors. So it's kind of all starting to come together.

And is there a whole bottle of things? Is it now too late?

No. No. Not this time. There's always time.

That would be so cool.

Yeah. I'll I'll add it to the list.

We'll have to

If it's not on the list, then basically it doesn't exist. But I should probably come up with a proposal for what it actually is. So, yeah, you tell you rolling along.

Fabulous. Is there anything else I sense there might be?

So we're getting super fast Internet. As opposed to the pretty slow Internet we've got at the moment.

So you're falling in the scope of the full fiber rollout.

Yeah. And Gwen's created a project to add Ethernet capability around the house.

Oh, that's an interesting one. So I've I've created a project to do it here in this office. Yeah. So up there is the sorry for the people who are just listening to this, but I'm pointing to the top of a very blue cupboard. And so that's where the BT full fibre modem, BT business hub three or whatever it is.

So it's made me

very happy.

In your idle moments, you can just do a speed test.

In my idle moments of which, you know, there are obviously many, I always do a speed test. Always. Else is worth doing with an idle moment.

First first thing in the morning, speed test.

Yes. Yes. I could do it now, but I'm not plugged in.

You'd just be disappointed.

No. No. I won't be disappointed. Right. So no. No. I won't be disappointed. But because Apple intercepts all my Internet traffic via iCloud

Uh-huh. And then in order to not process it. As a backdoor to the UK government.

Well, one likes to think not. Okay. So I'm I've gone to fast.com, which is Netflix's speed tester, and I can announce

This is so formal.

A speed test result over Wi Fi no less.

Of fast. Oh, no. That's the name of the site.

That is the name of the site. It could say one megabit.

It could do.

But it says something much closer to one gigabit. So for that speed test, I got 960 megabits per second. So it is insane. I have no idea what you could possibly do with that bandwidth. I mean, maybe if you get a room full of people to come and watch their own videos over it or something, it'd still be absurd.

In eight k?

Eight k. Yes. In eight k.

But yes. So our lives will be transformed. They will.

They will.

And all my various gaming machines will all have ethernet connections, which, to be fair, will be pretty good.

So have you ever heard anyone pronouncing it Ethernet?

You just know.

Let me ask you a different question. Have you ever heard anyone pronouncing it? Does everyone you know say Ethernet? Yeah. I've never heard anyone say

Ethernet. Okay.

To me, it's Ethernet. Maybe we should put a poll in the show note to definitively answer this question. I'm not

sure that would definitively answer

it. Are you saying we haven't got enough, of a sample size?

Well, yeah. Probably.

Well, that that we will have, though, because of all the testers in the world Yeah. Are gonna listen to this episode.

Is is it like sequel?

What? You mean SQL? SQL. Yeah. Yeah. Structured query language. Structured query

Should

yeah. Just refer to it by Sunday night.

When it's in trouble.

Structured query language. You come here right now. Explain this union. Explain yourself. Oh, dear.

You call that a left join.

There was somebody on Reddit who was talking about how SQL is a language because he's written 900 line statements in it Alright. To do something to do with extracting all the data from one database and putting it in a data warehouse or something. I was just sort of looking at it going, no. I never if I never see a 50 line SQL statement, then I I won't I won't mind.

Yeah. Yeah.

It'll be alright.

So just because you can do it in that Yeah. Doesn't mean that you you should.

Almost certainly Yeah. You shouldn't.

Yeah. Yeah. Call credit that the dot net developers used to say to the T SQL developers when they were in a room, they would say to the c T SQL developers, when will the real developers get here?

Oh, dear. Ouch. Oh, dear. Oh, dear.

Ouchy. Ouchy. Banter. Well, insults.

Yeah. Yeah. It's it's the kind of, thing that you call banter, but it's really just being rude.

Yeah. Exactly.

And offensive.

It's bant. Yeah. But, you know, the TC call that they wrote paid the wages. So

Hopefully, they got paid slightly more.

Yeah. Yeah. I think they did, you know. They kept it running.

Mind you, if they'd be, it'd be tempting to make up an absurd figure and say, well, that's why they pay me a hundred and 20,000 a year. Yeah. And then and then that would watch the people

watch the dot developers and

Yeah. They'll be running around indignantly complaining to their managers and things like that. It'd be quite sort of thing that would get you in trouble.

Yeah.

You're not allowed to discuss compensation. Oh, I wasn't. Clearly, I'm not compensated. Oh, dear. I can aspire.

So shall we move on to a a second thing?

Oh, we got two things.

We have two things.

That never happens. I'm just gonna open up the testing instead one again. Okay. Just, you know

Do do

do do do

that one again?

I'll just ask you the questions for that one. So, Ash, what do you think?

So my thing was is

It was, but but now

Sorry. I'm taking the podcast time machine too far there,

aren't I? Yes. Yes. You are. I think

you're allowed Thanks, everyone.

You're allowed to live in the moment of this even if it gets complicated sometimes.

So I I spotted an article on BBC News.

BBC News. Of all the A fine instrument of information.

Absolutely. Technology hotbeds of news.

Oh. It

was entitled Will Young Developers, not Will Young too.

We're all laughing about the the fact there's a person called Will Young.

It's not Will Young saying

Nobody younger than us has heard of Will Young. We should bring up Rick Astley and try and find a way of no. Let's not. Alright. Go on.

So will Young, developers, take on key open source software. That's the question.

You mean in in single combat?

Probably. Probably. So, basically, the maintainers of the various bits of open source software that commonly used. So in the article, they talk about curl.

As well, they might.

As well, they might. Basically, the the maintainers of cURL are saying, who's gonna take on my legacy when I when I don't want to do this anymore or when

I retire? Five years ago.

Yeah. Exactly. So it's like, who's gonna do this? And they're like, well, will any any younger developers take on take on the mantle? And I thought there was quite a few sort of interesting, like, threads

There are.

In in in the article.

That was young people these days.

Exactly. It did have that kind of tone.

It was

like, you know, you youngsters, you don't know what it was like. And that's literally like one of the first points of

They don't know they're born.

Yeah. Exactly.

We used to turn on a computer and a basic prompt would appear and you'd have to type in a program before you could do anything.

Yeah. And then get the get the punch cards and

Yeah.

Feed them through the and then realize you'd made a mistake

and then Took a tape, 300 board.

And then start again.

Start again.

And then the project manager would be like, what's going on?

And then bugs used to be actual bugs.

Yeah. Exactly.

Small insects trapped in your cause.

So, yeah. And that's one of the premises of the article, is that new developers

Of course it is.

They don't know what it was like before. So before open source, when Sun Microsystems owned everything or lots of things.

Well, to be fair, some market systems were generally quite generous with

Yeah. Sure.

The things.

Yeah. But to go back to, the days of closed source, I'm not sure quite sure we would actually go back there. But, basically, the the article posits and the people who will look after these these key libraries say because the next generation developers don't know what it was like to have a closed ecosystem for software development, then they don't value the open source enough.

So I think that's I mean, that's probably true. I mean, no human is wired to understand how things were Yeah. Before they were born. Is that there's that famous Douglas Adams quote, isn't there? He said, I've come up with a set of rules that describe our reactions to technologies.

He knew things.

He He knew things.

He did know things.

And it's kind of a bit like that. I mean, you can't expect people to remember that. Mhmm. But I wonder if a different question is how much they value it.

Yeah. Yeah. Because when it's ubiquitous, then you tend to not value it as much, do you?

Well, I mean, we have taps where we can swivel a little thing, and clean water comes out that we can drink. And that's it's like you would never

Good, delicious Yorkshire water.

Yorkshire water. With hardly any poop in it.

Acceptable levels of poop.

But the According to According to Yorkshire. Yeah. We're we're we're straying for quite You you go to the tap, you fill up your glass of water, and you drink it. Yeah. But there are some places in the world that you just can't do that. That's just not available. And clean water is much higher value because you kind of get it. And I think there's this gap there where that's probably not visible. There's a sort of economic thing as well. Yeah. 30 ago, software and technology was a good career

Mhmm.

In the sense that you you'd be well paid. You'd be paid more than a lot of other professions. True. And now it's a bit like we're just looking we're we're just going out looking for plankton. We're just gonna scraping all these developers up and then spitting them out again afterwards. Are

you are you likening technology companies to to Wales?

I love Wales.

Basking sharks.

I'm looking forward to going to Aberystwyth at some point over the next the next the coming months. No. But I I think I I think my feeling about it is that people have got less time now Yeah. And people are under more pressure. And one of the things you need for innovation is mental space.

Yeah.

And also, people made these. A lot of open source stuff just came from people scratching their own niches. Yeah. And you kind of struggle to come up with an itch for I want a piece of software that goes to a URL and retrieves the content because you've got curl. Yeah. And if you didn't have it, you've got wget. Yeah. It's almost like that itch is scratched. And what is the work of maintaining that software? It's fixing bugs

Yeah. And updating dependencies.

And packaging it. Yeah. That's not got the same joy in it

Yeah.

As making a tool that just does the exact thing you want. Yeah. And, you know, like Simon Willison's LLM and files to prompt tools, They're just scratching his own itch, and he's just publishing them. He's sharing them.

And saying, well, you might find these useful, but we will see, sort of thing. Yeah. Oh, but it wasn't designed to be open sourced. It was just designed probably came from a process, right, where it's like, oh, I want to do this thing. And then eventually, it's like, oh, other people might find this useful.

Yeah. Yeah. But it is interesting because the work of maintaining curl is just that. It's work. Yeah. And I'm going to say it might not be fun. Yeah. Yeah. I don't know. Maybe I'm being too harsh. Much more fun is making your own thing. Yeah. But then after a while, you're just getting rude people who take you for granted shouting at you because you haven't done the thing for them yet. Yeah. And I guess we've talked about the the kind of ability of this to be thankless as well.

Yeah. So The grim obligation.

Yeah. And so it's kind of I I think your BBC ask and asks a very good question. Thinking about it out loud like I was just doing has not led me to feel like, oh, yes. Well, here's the obvious answer.

Yeah. Yeah. Because the other thing as well is that a lot of these libraries are written in c, which is not top of the list for developers now, I don't think.

Well, they can have fun converting it into Rust, can't they? Yeah. Yeah. That's what you meant to do, isn't it?

Yeah. So It's the law. It's the rules. Linus, I made sure to say it right.

You did say it right. Yeah. The funny story about him sorry. I'm just gonna interrupt you or bring your point you're making. Somebody because he's been responsible for two huge innovations Yeah. In technology, and one of them is Linux and the other one is Git. And somebody asked him why did he call it Git, and he said, well, I am well known for making a a technology that I named after myself. I thought I'll just do it again. Fair. Fair enough. Sorry. Yes. Linus.

So Linus said within the projects that he's responsible for, it's okay to use Rust as a replacement language, which is is a good sign of the ability to adapt. Right? Although there was another contributor where this person likened, using Rust in some of these libraries, like cancer, which is particularly, like, tone deaf.

Yeah. It is, isn't it?

Really? But it's like you can't have it both ways. You can't retain your the purity that you that some open source developers may prioritize and welcome other people into the fold or new technologies to get other people into the fold. It's hard to do both of those things. Right?

I wonder if there's someone a la Linus who's going, right. I'm going to make an operating system in Rust.

No. Yeah. Just sort of someone someone somewhere who's a massive Rust enthusiast. He's just like, right. Yeah. It's time.

I've had enough of these stupid arguments.

Yeah. Yeah. Because they just seem a bit self defeating to me.

The people don't understand sometimes what makes other people want to do things. Yeah. I mean, Linus is famous for his bad tempered snotograms that he he writes to the colonel maintainer's mailing list.

Yeah. Yeah.

But I guess when you're him, you can kind of get away with that.

Mhmm.

But if I submit a pull request to curl, I'm I'd I have no, by the way, thought that the maintainer of curl is anything but a lovely person.

Yeah. Sure.

But if I submit the pull request to it to do something and was told not to be an idiot and to

Socratically questioned as to why you are so stupid.

Yes. Exactly. Thank you. You saved me from having to click the explicit flag in this episode. Then, you know, I might not I might think, well, okay. I was trying to help you, but apparently not in the right way. So you can do it on your own.

Yeah. Yeah. So I think in order to to get new people involved in open source maintenance, probably need to embrace newer technologies and People. People and have maybe a slightly more open attitude.

It's funny. Similar with Wikipedia editors. Mhmm. You hear about Wikipedia editors being quite savage to people who edit Wikipedia without knowing everything they know.

Yeah. Obviously, there are gatekeepers in open source software, aren't there? The people who can, you know, merge stuff in. Yeah. So, you know, you do have to sometimes it's a very narrow gate as well. Yeah. It is. Yeah. So,

And sometimes it has to be because you introduce something bad into the Linux kernel. The impact of that will be CrowdStrike sized at the very least Yeah. Potentially.

Yeah. Global apocalypse sized, depending on what it is.

Global apoca yes. What you said. Yeah. Disaster. Yeah. Global disaster science.

Yeah. So I think one of the things that struck me as well is, like, we talked a bit about jobs and time. So, like, good technology jobs can be hard to find or getting harder to find. So if you're a new developer and you're working as, like, an unpaid intern, which is a common pattern in

It's the devil's Yeah. Work that Do you

wanna do a further unpaid job for the love of programming? Probably not.

Well, I don't know. I mean, there is a a thing. There the love of programming is a thing. Yeah. And the love of making creatively making something is a thing. And so I think some people will want to do that, but not if they're knackered.

Yeah.

Not if they're exhausted because their job insists that they work stupid hours having magic time. I still haven't managed to invest that term with the level of contempt I actually feel for it in my voice, but I didn't know I tried. The thing is there is that love. It does exist. Yeah. And it makes me happy that it exists, But you can't take it for granted, I suppose. Yeah. And we take maintainers' open source software for granted because of the way it's been.

Yeah. And, you know, like I said, there there are benefits to being a maintainer as well, aren't there? I know a few people who've written a a library or something like that, got some consultancy out of it. You know, there's there's kind of indirect Yeah. Sort of benefits as well of doing it. But, again, it's like, well, you know, rather than budgeting it as, like, a grim community obligation that you should be thankful because you can't remember what it used to be like.

Eat your greens.

That you yeah. Exactly. So Amanda Brock, the CEO of Open UK, basically said, you've got a next generation who haven't engaged as a philanthropic community and volunteer community in the same way Wasted. At the same scale. Yeah. Yeah. Absolutely. So when I was reading that, basically, that makes me feel, of course, I'm not gonna contribute to open source now.

Thank you.

So statements like that always make me think, well

Screw you guys. Yeah. What we're doing here.

Pretty much. Pretty much. I don't know. I think my general sort of thought on the sort of questions raised by the article is, like, all the people involved in it, the current maintainers trying to attract new maintainers, the new developers coming through. There needs to be a bit of a like, they need to get a bit closer together in terms of what the offering is in order to do it rather than just the the the article to me badged it as, like, the grim obligation, and you should do it because, you know, your elders made sacrifices for you.

in there. Yeah. I mean, I I also thought that about it, and it was kind of just a bit finger wagging.

Yeah. Yeah.

And, really, we're talking about teams again.

Yeah. Yeah. So it's just like, how do you how do you then it's a different sort of team, isn't it? Because obviously extremely distributed.

Well, you know, there are a lot of things who've got volunteer teams in them. Yeah. I think that it's like that. Yeah. And when you've got people who are volunteers who are not there because of they've signed a contract or you're paying them

Yeah.

Then you have to work differently with them Yeah. Than you can do Yeah. If you're paying them.

Yeah. You can't really shout at volunteers and tell them to do things in a certain way because

Depends. Are you Gordon Ramsay?

Yeah. But I think on a kind of higher level away from, like, open source, I think sometimes articles like this are a bit doom mongery about people valuing, like, community and volunteering. Because actually loads of people volunteer for loads of things. They do. And you read an article like that, and it's like, well, no one's stepping forward to do this, which will be, like, demonstrable rubbish because there always will be.

Yeah. I think you're right.

Yeah. So, you know, it's not all doom and gloom. And I'm sure plenty of younger developers do contribute to open source. But Well,

it looks good on your CV if for no if Yeah. If nothing else.

Yeah. So, you know, maybe the current maintainers can do a little bit more to widen the net a little.

Yes. I think I think they could. Yeah. Also, though, once you get to a particular age, complaining about the younger generation these days is almost obligatory. Yeah.

Yeah. Absolutely.

I remember getting the bus as a kid. Do you ever did you get a bus when you were

Yeah.

A teenager or whatever? For particularly for school, I've got the bus to go to school Yeah. For some years. And I always remember hearing about seeing these old ladies complaining about the younger generation these days. They don't know this and that.

And before you know it, you all sat on the bus saying Yeah. But I'm right.

They were completely wrong.

So this time, it's for real.

Yeah. Yeah.

The next generation does not know they were born.

Yes. Yes. That's true. It's just our generation. We knew perfectly well that we were born.

Yeah. Exactly. So, you know, once the new generation of developers have made curl all broken and unmaintained with all dependencies and loads of bugs, then we'll see who's laughing.

Yes. Yes. Yes. We will.

Yeah. And then the finger wagging can begin.

Yes. Amid the collapse of the world's technological infrastructure. Oh, dear.

So that was my thing?

Well, I thought that was a monumental thing. A monumental. Because

Magnificent thing.

Marvelous, magnificent, and monumental, and many other great things that begin with them. Yep. Superb.

I actually went and looked up synonyms for magnificent that began with m. And? And there was loads, but I can't remember the French.

Oh, dear. So that's very annoying when you do that, and then you can't remember anything.

It was a very satisfying few minutes saying them to myself Yes. And then deeply unsatisfying, forgetting them.

Yes. Yes. We can put them in the next episode. Yeah. Show notes so we can remember them. We can

put them on the whiteboard. On the whiteboard? But then I won't be able to see them. So it's tough to

Yes. Tough to I will. I have my fun, so that's all that matters. So Super.

Super. So how do you get in touch with us at What A Lot of Things? But not not that's not the address.

Why would you do that?

How do you get in touch with, What A Lot of Things and the Oscar?

Well, you email [email protected]. Yeah. And so far, in our lives, some people have done that.

Yes. Indeed. Like, the Christmas party The

Christmas party we was a particular We bribed notable. We bribed people with access to a Christmas party, and they emailed us Yeah. To get it. So people will email us if we just make it worth their while.

I think it's that that a new generation of podcast listeners,

they just can't be bothered. They don't know they're born. Oh, dear. Oh, dear.

Oh, dear, indeed.

Well, on that note

Now we have chastised the new generation enough.

And we've dispensed with liberty and safety.

And We've decided that testing isn't dead.

It's still

Just on its

it's just in some difficulty. I keep hearing it wandering around saying, brains. Brains. I'm just trying that noise. That's testing.

Just trying to find a developer with a brain

So I could eat it.

So I can have a good conversation.

Testing is not dead. It's a zombie. You heard it here first.

Okay. Goodbye, Ian.

Goodbye, listener.

Goodbye, listener. Goodbye, Ian.

It's like goodnight, Boone, isn't it? Except not as not Oswald crafted. Good night, Ash.

Good night, Ian.