Im not dead yet (dunno about next week), for those that knew me here when I was around, but I really wanted to come to a place I know I could get some comments about it, but what the whole IT/Tech world right now?
Python and its CoC shenanigans
Linus leaving
Mozilla telemetry spying on you https://reddit.com/r/linux/...
And so on and on, the ride isnt over yet, right? (it never is, it only gets more fun from here on baby)

I already wrote a rant about this yesterday, but since I'm a sysadmin trying to convert to dev.. I dunno, maybe it's not a bad idea to muddy the waters a bit and talk about why not to be a sysadmin.

Personally I think it's that the perceived barrier to entry is just too high, while it isn't. You don't need a huge Ceph cluster and massive servers when you're just starting out. Why overbuild an appliance like that if it's gonna start out at maybe 5 requests a minute?

Let's take an example - DNS servers! So there's been this guy on the bind-users mailing list asking how to set up a DNS server on 2 public servers, along with a website. Nothing special I guess - you can read the thread here: https://0x0.st/ZY-d. Aside from the question being quite confusing, there was advice to read RFC's, get a book, read the BIND ARM, etc etc. And the person to deny this? No one less than Stephane Bortzmeyer, one of the people who works for nic.fr (so he maintains the .fr TLD) and wrote some of those RFC's as part of the DNSOP working group in the IETF. As for valid reasons to set up a DNS server? Could just be to learn how the DNS works, or hell even for fun. As far as professional DNS servers go.. this (https://0x0.st/ZYo9) is the nugget that powers the K root server, one of the 13 root servers that power the root zone of the internet, aka the zone apex. 2 RJ45 connections, and a console connection. The reason why this is possible is the massive recursor networks that ISP's, Google DNS, Cloudflare DNS, Quad9, etc etc provide. Point is, you don't need huge infrastructure to run a server!

Or maybe your business needs email. How many thousands of emails per second are you gonna need to build your mail server against? How many millions will you need to store? If your business has 10 employees and all of those manage about 10k emails total.. well that's easy, 100k emails total. Per second? Hundreds of emails per second per employee? Haha, of course not. Maybe you'll see an email a minute at most. That is not to say that all email services are like this - it is true that ISP's who offer email to their customers, and especially providers like Microsoft and Google do need massive mail servers that can handle thousands of emails per second. But you are not Microsoft or Google. So yeah, focus on the parts of email that are actually hard.. and there is plenty.

Among sysadmins you have this distinction between "professional" sysadmins and homelabbers. I don't mind the distinction itself but I think both augment each other. If you've started out by jumping into a heap of legacy at an established company, you will have plenty of resources, immediately high complexity, and probably a clusterfuck right away. But you will have massive amounts of resources. If you start out with a homelab, you will have not many resources, small workloads, and something completely new for you to build and learn with. And when running a server like that, you'll probably find that the resources required are quite small, to provide you with your new services. My DHCP servers take 12MB memory each. My DNS servers hover around the 40MB mark. The mail server.. to be fair that one consumes around 150. But if you'd hear the people saying that you need huge servers.. omg you need at least a TB of RAM on your server and 72 cores, massive disks and Ceph!1!

No you don't. All that does is scaring people away and creating a toxic environment for everyone. Stop it.

On C++ forum and see reference to Type Erasure (TE). Search around, some Java shit bleeding into other programming languages. Finally find an article that not only explains what TE is, but why you would use it in C++. ITS JUST FUCKING DUCK TYPING. Please stop using stupid names for stuff. You don't sound smarter. You sound like an asshole. Anyway, thinking about it does make sense to call it Type Erasure, but I still think it sounds pretentious. Cool concept, stupid name. Will continue to confuse people saying: "oh, you mean duck typing?"

Cool article:
https://davekilian.com/cpp-type-era...

The wikipedia article about TE doesn't explain shit about why you would even use it. Just repeats the same word salad of words I first saw about TE. I get that its jargon, but from the outside it just sounds like bullshit. I have never heard anyone I work with spew out shit like that. Even the ones with masters degrees in computer science.

I am not even sure I want to learn more about CS than what keeps me employed. I don't want to sound like this when I talk. I have already said shit in meetings about modern C++ that has colleagues (other sparkies, and some CS people) wondering what I was smoking. It wasn't even that jargony.

Don't mind me, just a sparky starting to understand why the CS world is so fucked. Maybe its just academia I can't stand. I dunno.

I should ask in a meeting if someone can define a monad for me.

!rant

For all of youse that ever wanted to try out Common Lisp and do not know where to start (but are interested in getting some knowledge of Common Lisp) I recommend two things:

As an introductory tutorial:
https://lisperati.com/casting.html/

And as your dev environment:
https://portacle.github.io/

Notice that the dev environment in question is Emacs, regardless of how you might feel about it as a text editor, i can recommend just going through the portacle help that gives you some basic starting points regarding editing. Learn about splitting buffers, evaluating the code you are typing in order for it to appear in the Common Lisp REPL (this one comes with an environment known as SLIME which is very popular in the Lisp world) as well as saving and editing your files.

Portacle is self contained inside of one single directory, so if you by any chance already have an Emacs environment then do not worry, Portacle will not touch any of that. I will admit that as far as I am concerned, Emacs will probably be the biggest hurdle for most people not used to it.
Can I use VS Code? Yes, yes you can, but I am not familiar with setting up a VSCode dev environment for Emacs, or any other environment hat comes close to the live environment that emacs provides for this?

Why the fuck should I try Common Lisp or any Lisp for that matter? You do not have to, I happen to like it a lot and have built applications at work with a different dialect of Lisp known as Clojure which runs in the JVM, do I recommend it? Yeah I do, I love functional programming, Clojure is pretty pure on that (not haskell level imo though, but I am not using Haskell for anything other than academic purposes) and with clojure you get the entire repertoire of Java libraries at your disposal. Moving to Clojure was cake coming from Common Lisp.

Why Common Lisp then if you used Clojure in prod? Mostly historical reasons, I want to just let people know that ANSI Common Lisp has a lot of good things going for it, I selected Clojure since I already knew what I needed from the JVM, and parallelism and concurrency are baked into Clojure, which was a priority. While I could have done the same thing in Common Lisp, I wanted to turn in a deliverable as quickly as possible rather than building the entire thing by myself which would have taken longer (had one week)

Am I getting something out of learning Common Lisp? Depends on you, I am not bringing about the whole "it opens your mind" deal with Lisp dialects as most other people do inside of the community, although I did experience new perspectives as to what programming and a programming language could do, and had fun doing it, maybe you will as well.

Does Lisp stands for Lots of Irritating Superfluous Parentheses or Los in stupid parentheses? Yes, also for Lost of Insidious Silly Parentheses and Lisp is Perfect, use paredit (comes with Portacle) also, Lisp stands for Lisp Is Perfect. None of that List Processing bs, any other definition will do.

Are there any other books? Yes, the famous online text Practical Common Lisp can be easily read online for free, I would recommend the Lisperati tutorial first to get a feel for it since PCL demands more tedious study. There is also Common Lisp a gentle introduction. If you want to go the Clojure route try Clojure for the brave and true.

What about Scheme and the Structure and Interpretation of Computer Programs? Too academic for my taste, and if in Common Lisp you have to do a lot of things on your own, Scheme is a whole other beast. Simple and beautiful really, but I go for practical in terms of Lisp, thus I prefer Common Lisp.

how did you start with Lisp?
I was stupid and thought I should start with it after a failed attempt at learning C++, then Java, and then Javascript when I started programming years ago. I was overwhelmed, but I continued. Then I moved to other things. But always kept Common Lisp close to heart. I am also heavy into A.I, Lisp has a history there and it is used in a lot of new and sort of unknown projects dealing with Knowledge Reasoning and representation. It is also Alien tech that contains many things that just seem super interesting to me such as treating code as data and data as code (back-quoting, macros etc)

I need some inspiration man......show me something? Sure, look for a game called Kandria in youtube, the creator, Shimera (Nicolas Hafner) is an absolute genius in the world of Lisp and a true inspiration. He coded the game in Common Lisp, he is also the person behind portacle. If that were not enough, he might very well also be Shirakumo, another prominent member of the Common Lisp Community.

Ok, you got me, what is the first thing in common lisp that I should try after I install the portacle environment? go to the repl and evaluate this:

(+ 0.1 0.2)

Watch in awe at what you get.

In the truest and original sense of the phrase (MIT based) "happy hacking!"

I had the idea that part of the problem of NN and ML research is we all use the same standard loss and nonlinear functions. In theory most NN architectures are universal aproximators. But theres a big gap between symbolic and numeric computation.

But some of our bigger leaps in improvement weren't just from new architectures, but entire new approaches to how data is transformed, and how we calculate loss, for example KL divergence.

And it occured to me all we really need is training/test/validation data and with the right approach we can let the system discover the architecture (been done before), but also the nonlinear and loss functions itself, and see what pops out the other side as a result.

If a network can instrument its own code as it were, maybe it'd find new and useful nonlinear functions and losses. Networks wouldn't just specificy a conv layer here, or a maxpool there, but derive implementations of these all on their own.

More importantly with a little pruning, we could even use successful examples for bootstrapping smaller more efficient algorithms, all within the graph itself, and use genetic algorithms to mix and match nodes at training time to discover what works or doesn't, or do training, testing, and validation in batches, to anneal a network in the correct direction.

By generating variations of successful nodes and graphs, and using substitution, we can use comparison to minimize error (for some measure of error over accuracy and precision), and select the best graph variations, without strictly having to do much point mutation within any given node, minimizing deleterious effects, sort of like how gene expression leads to unexpected but fitness-improving results for an entire organism, while point-mutations typically cause disease.

It might seem like this wouldn't work out the gate, just on the basis of intuition, but I think the benefit of working through node substitutions or entire subgraph substitution, is that we can check test/validation loss before training is even complete.

If we train a network to specify a known loss, we can even have that evaluate the networks themselves, and run variations on our network loss node to find better losses during training time, and at some point let nodes refer to these same loss calculation graphs, within themselves, switching between them dynamically..via variation and substitution.

I could even invision probabilistic lists of jump addresses, or mappings of value ranges to jump addresses, or having await() style opcodes on some nodes that upon being encountered, queue-up ticks from upstream nodes whose calculations the await()ed node relies on, to do things like emergent convolution.

I've written all the classes and started on the interpreter itself, just a few things that need fleshed out now.

Heres my shitty little partial sketch of the opcodes and ideas.
https://pastebin.com/5yDTaApS

I think I'll teach it to do convolution, color recognition, maybe try mnist, or teach it step by step how to do sequence masking and prediction, dunno yet.

disclaimer i dont understand css to begin with so you can discard my opinion

You have all these options for width https://developer.mozilla.org/en-US... , but guess what none of them do anything different as you brute force try them all in the chrome debugger. Dunno what cascades except my butthurt

so fuck it ~1000% width works and has an ugly overhang, but fuck front end