@Devintrix , congrats and happy lifes with your wife. this joke is for you :)

Dear Tech Support:

Last year I upgraded from Girlfriend 7.0 to Wife 1.0 and noticed that the new program began running unexpected child processing that took up a lot of space and valuable resources.

In addition, Wife 1.0 installs itself into all other programs and launches during system initialization, where it monitors all other system activity. Applications such as PokerNight 10.3, Drunken Boys Night 2.5 and Monday Night football 5.0 no longer run, crashing the system whenever selected.

I cannot seem to keep Wife 1.0 in the background while attempting to run some of my other favorite applications. I am thinking about going back to Girlfriend 7.0, but un-install does not work on this program.

Can you help me please?

Thanks,

Joe

——————————————————–
Dear Joe:

This is a very common problem men complain about but is mostly due to a primary misconception. Many people upgrade from Girlfriend 7.0 to Wife 1.0 with the idea that Wife 1.0 is merely a “UTILITIES & ENTERTAINMENT” program. Wife 1.0 is an OPERATING SYSTEM and designed by its creator to run everything.

It is unlikely you would be able to purge Wife 1.0 and still convert back to Girlfriend 7.0. Hidden operating files within your system would cause Girlfriend 7.0 to emulate Wife 1.0 so nothing is gained.

It is impossible to un-install, delete, or purge the program files from the system once installed. You cannot go back to Girlfriend 7.0 because Wife 1.0 is not designed to do this. Some have tried to install Girlfriend 8.0 or Wife 2.0 but end up with more problems than the original system.

I recommend you keep Wife 1.0 and just deal with the situation. Having Wife 1.0 installed myself, I might also suggest you read the entire section regarding General Partnership Faults (GPFs). You must assume all responsibility for faults and problems that might occur, regardless of their cause. The best course of action will be to enter the command C:\APOLOGIZE. The system will run smoothly as long as you take the blame for all the GPFs.

Wife 1.0 is a great program, but very high maintenance. Consider buying additional software to improve the performance of Wife 1.0. I recommend Flowers 2.1, Jewelry 2.2, and Chocolates 5.0.

Do not, under any circumstances, install Secretary With Short Skirt 3.3. This is not a supported application for Wife 1.0 and is likely to cause irreversible damage to the operating system.

Best of luck,
Tech Support

Story time:
Yesterday I wanted to go to the theater with my girlfriend. It was her idea because as a student you can get reduced tickets for the play, but only via the online store exactely two hours before the play starts. We had already tried two weeks before but with no success. So this time I said i want to be on my pc with a proper browser and not a mobile version like last time. So we are sitting at home me in front of their website on one screen and with a clock on the other screen. Two minutes realy i hit refresh and I get a selection for the reduced tickets, nice.

You would think.

After selecting the amount. ERROR: Can not get your tickets. I was like fuck they are already sold out because it's a popular play. But hey let's try again. I got one ticket but not the second one, okay strange lets try again, same ERROR again. WHAT the FUCK, no feedback what so ever. My girlfriend had then the idea that they maybe restricted the amount for reduced tickets to one (does not state this explicitly but hey lets give it a shot). Use second browser select one ticket. ERROR can not get you the amount of seats. Rage level near to a 1000 why did it work two minutes before but not anymore. Trying around for five more minutes finally got the second ticket.

Now the real fun begins.

Proceeding to checkout should not be that hard you would think, but you need to be registered for that. Okay so let's do that. The salutation is not required neither is the address for the tickets but you need to have a company name??!!!!! The fuck?? I am not self employed and neither are a most other people around here so why is this field mandatory? Beeing a little under stress I decided to found the "asdf" company with my girlfriend.

Now one would think checking out is easy. Not so fast.

After accepting the terms of service another ERROR, unable to accept your data. What data? I did not input anything new? Where does this come from? Ok never mind I am going to pay with credid card that must work!
ERROR: Internal paymentservice initialization failure! Sorry what? I thought maybe I was to long idle in this browser and they do not reserve the tickets for so long (which would be no surprise to me at this point). Let's try again. Nope same error.

Now my rage level was really over 9000 but we really wanted to go so I decided to call the customer SUPPORT. Or better to say I had a answering maching telling me for ten minutes how sorry they are that this takes so long, yeah you bet. Then and this is now really great: the support guy asks me: "What error do you see? Internal paymentservice initialization failure?" I was like, okay he knows this so they need to know how to handle it. FUCK NO. "Sorry I can't help you. This is our payment system maybe they (IT) are doing some maintenance I can't halp you. Call the theater directly good day." Sorry what just happened, you fuckers are the vendors for the tickets for nearly all big events around here and the theater explicitly states to call you for tickets but you can not help me? Like hell.

This process took 25 very frustrating minutes and I was really angry and wanted to quit, then I saw that there is also a paypal option which I had not tried. With very little hope i selected everything for the payment, registered with paypal and they told me I already had an account. So reactivated this five year old account payed with all the mobile passwords and tans to finally, after 30 fucking minutes, get a pdf file for a ticket. Repeated the last step for the second ticket and with some time left to get there we were off.

Right, you pesky type initialization bug, I'm gonna find you. Hiding by throwing your exception in an external library won't help you. I'll download the sources, library by library, and look for you there too.

I *will* track you down, bastard, and exterminate you locally. Then, I'll make a pull request and kick your sorry ass off GitHub, off the internet and off the bloody surface of the earth. Oh, you have no idea how dead you're going to be when I'm done here!

WordPress related, get ready for some disgust.

So today early in the morning my boss forwarded me an email from a client, it was about a bug, and asked me if I can have a look at it and fix it.
"Yaay, WordPress!" I thought and opened the page containing the mentioned bug. She wrote that in the italian version of the page, users can select dates in the calendar, which should be disabled, like in the german version.

So yeah, I opened the code. Everything in the function looked perfect. Really. And the Data was also correctly set in the backend of WP.

The function was only 3 lines of code:

- Get the german post ID of the current post (german or italian) by its ID (using a Polylang function)

- Get an Advanced Custom Fields field by name and from a post with the ID from before

- json_encode its content and echo it to a JS var for initialization and later use in some AngularJS.

No fucking missing semicolon, it was fucking perfect like a sunset with your soulmate.

So I tried to find the bug with my personal way of debugging:

"Shitstream Debugging"
When a creek suddenly is full of water mixed with shit, walk upstream through the turds until you reach clear water. This is where the bug is.

=> So I first looked at the HTML source: Turds.
=> Then the ACF field content: Still turds.
=> Then the ID of the german post: Shit stain and turds (var_dump: null)
=> Please god at least $post->ID? Nope, fart smell and turds.
=> Nothing more to check: Clear fucking water and the flowery smell of 99 devVirgins

So it replaced $post->IT with get_the_ID() and it worked like a charm.

Afterwards I feel stupid, but $post->IT worked all the times before...

Conclusion:
FUCK YOU WORDPRESS YOU UGLY PIECE OF HUMAN-CENTIPEDE-PROCESSED-DOGFART.

Thanks for your patience.
Only one beer was sucked dry during the writing of this fucking rant.

Time to rant about JavaScript tutorials.

If you don't know the 'jQuery basic arithmetic' joke, Google it now. It'll make you laugh, promised.

In that manner i just remembered a JavaScript tutorial my fiancee tried to follow when she did an internship at the company i work for last year.

She was tasked to create a temperature interface for our server rack, which she wanted to do via an Arduino and a webserver aswell as an SQL database.

The Arduino part wasn't really a problem, but since she had no experience with js she very closely clinged to a chart visualisation tutorial.

All of that worked very well, but beeing the person i am i looked at the code and found something off.

The chart library had no dependencies to external libraries or any local files for any of them. Though the tutorial used a jQuery import.

So why did it use jQuery?
Well...
To load the chart initialization after the page has loaded.

So they pulled the entirety of jQuery in just to do what fucking window.addEventListener('DOMContentLoaded',function(){...}); could have done.

I wonder how many people who just want something to work did this shit. I hate it that so many tutorials do not adhere any kinds of standards, override behavior because they don't like it, even though it may have a very good reason to exist, pull entire libraries in for something vanilla <language> can do in 3 lines, etc.

Fuck.

Touch Initialization failed!
No train Ticket for me :(

2 things I'm working on now:

#1 a personal project I am hoping to commercialize and turn it into my moneymaker. Hoping it'd at least be enough to pay the bills and put food on my table so I could forget 9/5 for good. But it has a potential of becoming a much, MUCH bigger thing. This would need the right twist tho, and I'm not sure if I am "the right twister" :) We'll see.

#2 smth I'm thinking of opensourcing once finished -- a new form of TLS. This model could be unbreakable by even quantum computing once it's mature enough to crack conventional TLS. I'm probably gonna use md5 or smth even weakier - I'm leveraging the weakness of hashing functions to make my tool stronger :)

I mean how long can we be racing with more powerful computers, eh? Why not use our weakneses to make them our strengths?

Unittests are already passing, I just haven't polished all the corner-cases and haven't worked out a small piece of the initialization process yet. But it's very close

Worst thing you've seen another dev do? Here is another.

Early into our eCommerce venture, we experienced the normal growing pains.

Part of the learning process was realizing in web development, you should only access data resources on an as-needed basis.

One business object on it's creation would populate db lookups, initialize business rule engines (calling the db), etc.

Initially, this design was fine, no one noticed anything until business started to grow and started to cause problems in other systems (classic scaling problems)

VP wanted a review of the code and recommendations before throwing hardware at the problem (which they already started to do).

Over a month, I started making some aggressive changes by streamlining SQL, moving initialization, and refactoring like a mad man.

Over all page loads were not really affected, but the back-end resources were almost back to pre-eCommerce levels.

The main web developer at the time was not amused and fought my changes as much as she could.

Couple months later the CEO was speaking to everyone about his experience at a trade show when another CEO was complementing him on the changes to our web site.
The site was must faster, pages loaded without any glitches, checkout actually worked the first time, etc.
CEO wanted to thank everyone involved etc..and so on.

About a week later the VP handed out 'Thank You' certificates for the entire web team (only 4 at the time, I was on another team). I was noticeably excluded (not that I cared about a stupid piece of paper, but they also got a pizza lunch...I was much more pissed about that). My boss went to find out what was going on.

MyBoss: "Well, turned out 'Sally' did make all the web site performance improvements."
Me: "Where have you been the past 3 months? 'Sally' is the one who fought all my improvements. All my improvements are still in the production code."
MyBoss: "I'm just the messenger. What would you like me to do? I can buy you a pizza if you want. The team already reviewed the code and they are the ones who gave her the credit."
Me: "That's crap. My comments are all over that code base. I put my initials, date, what I did, why, and what was improved. I put the actual performance improvement numbers in the code!"
MyBoss: "Yea? Weird. That is what 'Tom' said why 'Sally' was put in for a promotion. For her due diligence for documenting the improvements."
Me:"What!? No. Look...lets look at the code"

Open up the file...there it was...*her* initials...the date, what changed, performance improvement numbers, etc.

WTF!

I opened version control and saw that she made one change, the day *after* the CEO thanked everyone and replaced my initials with hers.

She knew the other devs would only look at the current code to see who made the improvements (not bother to look at the code-differences)

MyBoss: "Wow...that's dirty. Best to move on and forget about it. Let them have their little party. Let us grown ups keeping doing the important things."

Does most memorable in a bad way count? 🤔

He left almost 2 years ago..or even more.. left a bunch of bugs and logical fuckups for me to fix.. some already fixed, some still lingering there..

I want to not blame him for everything, since we lack proper code review protocols and all.. but I've asked on several occasions if he understands the problem and what must be done..and the answer was always yes..results, after I got time to check up on him, the code he wrote was most probably copy pasted from stack overflow or somewhere else.. butchered in any and every way possible..

And of course already checked in to TFS.. along with bunch of files that were not even changed (he didn't bother to check that and exclude them) + a bunch of files from other projects... Told him to not do that on several occasions too, but he still managed to fuck things up this way.. leaving all the uncommented debugger; crap and alerts in the js files..

On one occasion I was working on new GUI..api part I already finished..got the order from above to delegate this to him as it is not much he can fuck up so I could focus on more important & complex stuff..
Maybe additional 4h of work + testing for everything..
I show him the prepared files, one controller, one view..explained what parts of code goes where etc.. a little short of writing everything myself.. Ask if he understands what needs to be done & how and told him if he has any problems/questions to ask me asap..
Said he understood what needs to be done.. after a day or two he asks me why something is not workig as expected.. I check the files, correct initialization was commented out and all the code was stuffed in the view file.. Took him another day to move the code to proper files.. Not sure about the possible bugs left there as the client later decided that they will not be using this..

I later found out that years of C# experience on his CV was actually a school course.. he didn't even know why the changes on api are not showing up..because he didn't know that he has to build the code..

I mean, if he was honest when asked about experience with .net, we would've taken a month or two to just explain everything from the start..

But as he didn't and based on his CV (much more experience with .net than me) and 'I understand everything' attitude from the start I assumed he knows WTF was he doing..

Boy was I wrong..

He was also more interested in how much I get payed and if I have a company phone etc..than actually doing his job.. I fucking hate chit chat, and this..well.. he didn't get the hints that this is in no way appropriate to ask.. I've told him that if he has problems with his pay and bonuses that he should talk to the management and not me about this..and that I'm only interested in his actual work and progress..

So yeah, I'll definitely be remembering this guy till the day I die..

Me: Startup initialization tests fail so the system won't boot properly it until we fix those issues.

Manager: comment out the lines of code that are causing problems.

🤔 So we are just going to feed power to some hardware and see what happens I guess

In 2015 I sent an email to Google labs describing how pareidolia could be implemented algorithmically.
The basis is that a noise function put through a discriminator, could be used to train a generative function.
And now we have transformers.

I also told them if they looked back at the research they would very likely discover that dendrites were analog hubs, not just individual switches. Thats turned out to be true to.

I wrote to them in an email as far back as 2009 that attention was an under-researched topic. In 2017 someone finally got around to writing "attention is all you need."

I wrote that there were very likely basic correlates in the human brain for things like numbers, and simple concepts like color, shape, and basic relationships, that the brain used to bootstrap learning. We found out years later based on research, that this is the case.

I wrote almost a decade ago that personality systems were a means that genes could use to value-seek for efficient behaviors in unknowable environments, a form of adaption. We later found out that is probably true as well.

I came up with the "winning lottery ticket" hypothesis back in 2011, for why certain subgraphs of networks seemed to naturally learn faster than others. I didn't call it that though, it was just a question that arose because of all the "architecture thrashing" I saw in the research, why there were apparent large or marginal gains in slightly different architectures, when we had an explosion of different approaches. It seemed to me the most important difference between countless architectures, was initialization.

This thinking flowed naturally from some ideas about network sparsity (namely that it made no sense that networks should be fully connected, and we could probably train networks by intentionally dropping connections).

All the way back in 2007 I thought this was comparable to masking inputs in training, or a bottleneck architecture, though I didn't think to put an encoder and decoder back to back.

Nevertheless it goes to show, if you follow research real closely, how much low hanging fruit is actually out there to be discovered and worked on.

And to this day, google never fucking once got back to me.

I wonder if anyone ever actually read those emails...

Wait till they figure out "attention is all you need" isn't actually all you need.

p.s. something I read recently got me thinking. Decoders can also be viewed as resolving a manifold closer to an ideal form for some joint distribution. Think of it like your data as points on a balloon (the output of the bottleneck), and decoding as the process of expanding the balloon. In absolute terms, as the balloon expands, your points grow apart, but as long as the datapoints are not uniformly distributed, then *some* points will grow closer together *relatively* even as the surface expands and pushes points apart in the absolute.

In other words, for some symmetry, the encoder and bottleneck introduces an isotropy, and this step also happens to tease out anisotropy, information that was missed or produced by the encoder, which is distortions introduced by the architecture/approach, features of the data that got passed on through the bottleneck, or essentially hidden features.

In the 90s most people had touched grass, but few touched a computer.

In the 2090s most people will have touched a computer, but not grass.

But at least we'll have fully sentient dildos armed with laser guns to mildly stimulate our mandatory attached cyber-clits, or alternatively annihilate thought criminals.

In other news my prime generator has exhaustively been checked against, all primes from 5 to 1 million. I used miller-rabin with k=40 to confirm the results.

The set the generator creates is the join of the quasi-lucas carmichael numbers, the carmichael numbers, and the primes. So after I generated a number I just had to treat those numbers as 'pollutants' and filter them out, which was dead simple.

Whats left after filtering, is strictly the primes.

I also tested it randomly on 50-55 bit primes, and it always returned true, but that range hasn't been fully tested so far because it takes 9-12 seconds per number at that point.

I was expecting maybe a few failures by my generator. So what I did was I wrote a function, genMillerTest(), and all it does is take some number n, returns the next prime after it (using my functions nextPrime() and isPrime()), and then tests it against miller-rabin. If miller returns false, then I add the result to a list. And then I check *those* results by hand (because miller can occasionally return false positives, though I'm not familiar enough with the math to know how often).

Well, imagine my surprise when I had zero false positives.

Which means either my code is generating the same exact set as miller (under some very large value of n), or the chance of miller (at k=40 tests) returning a false positive is vanishingly small.

My next steps should be to parallelize the checking process, and set up my other desktop to run those tests continuously.
Concurrently I should work on figuring out why my slowest primality tests (theres six of them, though I think I can eliminate two) are so slow and if I can better estimate or derive a pattern that allows faster results by better initialization of the variables used by these tests.

I already wrote some cases to output which tests most frequently succeeded (if any of them pass, then the number isn't prime), and therefore could cut short the primality test of a number. I rewrote the function to put those tests in order from most likely to least likely.

I'm also thinking that there may be some clues for faster computation in other bases, or perhaps in binary, or inspecting the patterns of values in the natural logs of non-primes versus primes. Or even looking into the *execution* time of numbers that successfully pass as prime versus ones that don't. Theres a bevy of possible approaches.

The entire process for the first 1_000_000 numbers, ran 1621.28 seconds, or just shy of a tenth of a second per test but I'm sure thats biased toward the head of the list.

If theres any other approach or ideas I may be overlooking, I wouldn't know where to begin.

I can't stand Swift's initializers. No other languages have the problem with constructors/initializers that Swift does. It's a complete failure of a feature and to hell with safety if it comes with this cost.

Just to illustrate how ridiculous it gets, I want to have a class where my initialization logic can be split among reusable parts. That is, the logic that initializes the class with no parameters has logic that I want to reuse in my other initializers. Simple DRY stuff. Well, the only way I can do that in Swift is if I use a convenience initializer that calls another one. But convenience initializers have completely different rules from designated initializers (again, something only Swift does).

For example, you can't access "self" until you call a designated initializer. You can't chain designated initializers, and if you want to chain anything in the same class you have to handcuff yourself by using a "convenience" initializer (there's nothing convenient about them, I might add).

So now I want to subclass my class and initialize myself using one of my superclass initializers. Oh but the one I want to call is a *convenience* initializer so I can't, unless I turn my new initializer into a convenience initializer. Except wait, a convenience initializer must delegate with self.init(), so it can't even call a superclass initializer!

And it just goes round and round and round. I don't know if I should try to convert all of my initializers to convenience initializers or the other way around.

Why all this nonsensical madness? Get rid of the distinction and go back to nice clean powerful initializers like Objective-C. I mean what does it have to take? This is a complete nightmare.

When I was an apprentice in a small company, ...

my boss told me that his company would never ship release builds, because the "evil optimization option" is responsible for breaking his code.
My first thought was that it wouldn't make any sense at all. The default option for code optimization is always set to zero.
After investigating his code, I found out that he didn't care to properly initialize his variables. The default compiler option for debug builds did implicitly initialize all variables to zero. After that I've confronted him with the fact that implicit null initialization does not conform to the standard of C and C++. He didn't believe me what I was saying and he was questioning my knowledge about C and C++. He refused to fix his code to this day, so he keeps building his libraries and applications always in debug mode.
Bonus fact: He would never build 64-bit applications, because his serialization functions do get incompatible with exisiting file formats.

O.M.F.G how much longer is that gonna take?!?! It's been 2 days already! 2 fucking days!!!

That's it. Fuck the docs. I'm going hardcore now: hooking up a debugger and diving into the unseen depths of Spring's automatic JPA repo initialization.

Unless any one of you knows how to
1. Manually create jpa repo interface bean
2. Make transactions work [also enable them manually by declaring beans in @Configuration]

and do all this in a single @Bean method. Jpa provider - hibernate.

Nobody? Not a single one? Allright then, I'm going in :(

This is gonna be a long post, and inevitably DR will mutilate my line breaks, so bear with me.
Also I cut out a bunch because the length was overlimit, so I'll post the second half later.

I'm annoyed because it appears the current stablediffusion trend has thrown the baby out with the bath water. I'll explain that in a moment.

As you all know I like to make extraordinary claims with little proof, sometimes
for shits and giggles, and sometimes because I'm just delusional apparently.

One of my legit 'claims to fame' is, on the theoretical level, I predicted
most of the developments in AI over the last 10+ years, down to key insights.
I've never had the math background for it, but I understood the ideas I
was working with at a conceptual level. Part of this flowed from powering
through literal (god I hate that word) hundreds of research papers a year, because I'm an obsessive like that. And I had to power through them, because
a lot of the technical low-level details were beyond my reach, but architecturally
I started to see a lot of patterns, and begin to grasp the general thrust
of where research and development *needed* to go.

In any case, I'm looking at stablediffusion and what occurs to me is that we've almost entirely thrown out GANs. As some or most of you may know, a GAN is
where networks compete, one to generate outputs that look real, another
to discern which is real, and by the process of competition, improve the ability
to generate a convincing fake, and to discern one. Imagine a self-sharpening knife and you get the idea.

Well, when we went to the diffusion method, upscaling noise (essentially a form of controlled pareidolia using autoencoders over seq2seq models) we threw out
GANs.

We also threw out online learning. The models only grow on the backend.
This doesn't help anyone but those corporations that have massive funding
to create and train models. They get to decide how the models 'think', what their
biases are, and what topics or subjects they cover. This is no good long run,
but thats more of an ideological argument. Thats not the real problem.
The problem is they've once again gimped the research, chosen a suboptimal
trap for the direction of development.

What interested me early on in the lottery ticket theory was the implications.
The lottery ticket theory says that, part of the reason *some* RANDOM initializations of a network train/predict better than others, is essentially
down to a small pool of subgraphs that happened, by pure luck, to chance on
initialization that just so happened to be the right 'lottery numbers' as it were, for training quickly.

The first implication of this, is that the bigger a network therefore, the greater the chance of these lucky subgraphs occurring. Whether the density grows
faster than the density of the 'unlucky' or average subgraphs, is another matter.
From this though, they realized what they could do was search out these subgraphs, and prune many of the worst or average performing neighbor graphs, without meaningful loss in model performance. Essentially they could *shrink down* things like chatGPT and BERT.

The second implication was more sublte and overlooked, and still is.
The existence of lucky subnetworks might suggest nothing additional--In which case the implication is that *any* subnet could *technically*, by transfer learning, be 'lucky' and train fast or be particularly good for some unknown task.

INSTEAD however, what has happened is we haven't really seen that. What this means is actually pretty startling. It has two possible implications, either of which will have significant outcomes on the research sooner or later:
1. there is an 'island' of network size, beyond what we've currently achieved,
where networks that are currently state of the3 art at some things, rapidly converge to state-of-the-art *generalists* in nearly *all* task, regardless of input. What this would look like at first, is a gradual drop off in gains of the current approach, characterized as a potential new "ai winter", or a "limit to the current approach", which wouldn't actually be the limit, but a saddle point in its utility across domains and its intelligence (for some measure and definition of 'intelligence').

- int varX;
v/s
- int varX = 0;

makes a shit ton of difference 😖😖😖

Always remember to initialize.

Got VS running, SDL up and running and outputting, and angelscript included. Only getting linker errors on angel at the moment, not on inclusion, but on calling engine initialization.
Who knows what it is. Devs recommended precompiling but I wanted to compile with the project rather than as a dll (maybe I'm doing something stupid though, too new to know).

Goal is to do for sdl, cpp, and angelscript, what LOVE2d did for lua. Maybe half baked, and more just an experiment to learn and see if I can.

Would be cool to script in cpp without having to fuck with compilers and IDEs.

As simple as 1. write c++, 2. script is compiled on load, 3. have immediate access to sdl in the same language that the documentation and core bindings are written for.

Maybe make something a little more batteries-included than what lua and love offer out of the box, barebones editors and tooling and the like, but thats off in the near future and just a notion rather than a solid plan.

Needed to take a break from coding my game and here I am..experimenting with more code.

Something is wrong with me.

Our Swift teacher at college manually creates a Podfile for every project and copies and pastes the basic initialization from an existing project, pastes the cocoapod dependencies and then installs them using terminal instead of just doing a pod init and then using nano or vim to paste the dependencies right inside the terminal. These are the times I genuinely feel sad for the way Indian education system is and the way we're taught coding in colleges here.

They keep training bigger language models (GPT et al). All the resear4chers appear to be doing this as a first step, and then running self-learning. The way they do this is train a smaller network, using the bigger network as a teacher. Another way of doing this is dropping some parameters and nodes and testing the performance of the network to see if the smaller version performs roughly the same, on the theory that there are some initialization and configurations that start out, just by happenstance, to be efficient (like finding a "winning lottery ticket").

My question is why aren't they running these two procedures *during* training and validation?

If [x] is a good initialization or larger network and [y] is a smaller network, then
after each training and validation, we run it against a potential [y]. If the result is acceptable and [y] is a good substitute, y becomes x, and we repeat the entire procedure.

The idea is not to look to optimize mere training and validation loss, but to bootstrap a sort of meta-loss that exists across the whole span of training, amortizing the loss function.

Anyone seen this in the wild yet?

Boss wanted our AES implemented to use an Initialization Vector. I changed the implementation to add an IV and all the tests passed on the first try!

I changed the implementation to fail just to make sure I wasn't getting false positives.

I have, once again, figured out why I keep dropping C as a language. The answer is because the errors are incredibly unhelpful and actively want to implode your brain.

Examples currently being spat out by gcc for my driver:

`error: conflicting types for ‘block_read’; have ‘ssize_t(struct file *, char *, size_t, loff_t *)’ {aka ‘int(struct file *, char *, unsigned int, long long int *)’}`
`note: previous declaration of ‘block_read’ with type ‘ssize_t(struct file *, char *, size_t, loff_t *)’ {aka ‘int(struct file *, char *, unsigned int, long long int *)’}`

`error: initialization of ‘ssize_t (*)(struct file *, char *, size_t, loff_t *)’ {aka ‘int (*)(struct file *, char *, unsigned int, long long int *)’} from incompatible pointer type ‘ssize_t (*)(struct file *, char *, size_t, loff_t *)’ {aka ‘int (*)(struct file *, char *, unsigned int, long long int *)’}`

Go character by character for those types as listed, and tell me where they differ, because I can't find it.

Who thought this was a good idea

oh well if they couldn't care to initialize the variable I GUESS WE'LL MAKE IT RANDOM

There is probably a reason for it... somewhere

Sometimes in our personal projects we write crazy commit messages. I'll post mine because its a weekend and I hope someone has a well deserved start. Feel free to post yours, regex out your username, time and hash and paste chronologically. ISSA THREAD MY DUDES AND DUDETTES
--
Initialization of NDM in Kotlin
Small changes, wiping drive
Small changes, wiping drive
Lottie, Backdrop contrast and logging in implementation
Added Lotties, added Link variable to Database Manifest
Fixed menu engine, added Smart adapter, indexing, Extra menus on home and Calendar
b4 work
Added branch and few changes
really before work
Merge remote-tracking branch 'origin/master'
really before work 4 sho
Refined Search response
Added Swipe to menus and nested tabs
Added custom tab library
tabs and shh
MORE TIME WASTED ON just 3 files
api and rx
New models new handlers, new static leaky objects xd, a few icons
minor changes
minor changesqwqaweqweweqwe
db db dbbb
Added Reading display and delete function
tryin to add web socket...fail
tryin to add web socket...success
New robust content handler, linked to a web socket. :) happy data-ring lol
A lot of changes, no time to explain
minor fixes ehehhe
Added args and content builder to content id
Converted some fragments into NDMListFragments
dsa
MAjor BiG ChANgEs added Listable interface added refresh and online cache added many stuff
MAjor mAjOr BiG ChANgEs added multiClick block added in-fragment Menu (and handling) added in-fragment list irem click handling
Unformatted some code, added midi handler, new menus, added manifest
Update and Insert (upsert) extension to Listable ArrayList
Test for hymnbook offline changing
Changed menuId from int to key string :) added refresh ...global... :(
Added Scale Gesture Listener
Changed Font and size of titlebar, text selection arg. NEW NEW Readings layout.
minor fix on duplicate readings
added isUserDatabase attribute to hymn database file added markwon to stanza views
Home changes :)
Modular hymn Editing
Home changes :) part 2
Home changes :) part 3
Unified Stanza view
Perfected stanza sharing
Added Summernote!!
minor changes
Another change but from source tree :)))
Added Span Saving
Added Working Quick Access
Added a caption system, well text captions only
Added Stanza view modes...quite stable though
From work changes
JUST a [ush
Touch horizontal needs fix
Return api heruko
Added bible index
Added new settings file
Added settings and new icons
Minor changes to settings
Restored ping
Toggles and Pickers in settings
Added Section Title
Added Publishing Access Panel
Added Some new color changes on restart. When am I going to be tired of adding files :)
Before the confession
Theme Adaptation to views
Before Realm DB
Theme Activity :)
Changes to theme Activity
Changes to theme Activity part 2 mini
Some laptop changes, so you wont know what changed :)
Images...
Rush ourd
Added palette from images
Added lastModified filter
Problem with cache response
works work
Some Improvements, changed calendar recycle view
Tonic Sol-fa Screen Added
Merge Pull
Yes colors
Before leasing out to testers
Working but unformated table
Added Seperators but we have a glithchchchc
Tonic sol-fa nice, dots left, and some extras :)))
Just a nice commit on a good friday.
Just a quickie
I dont know what im committing...

I finally got the lstm to a training and validation loss of < 0.05 for predicting the digits of a semiprime's factors.

I used selu activation with lecun normal initialization on a dense decoder, and compiled the model with Adam as the optimizer using mean squared error.

Selu is self-normalizing, meaning it tends to mean 0 and preserves a standard deviation of one, so it eliminates the exploding/vanishing gradient problem. And I can get away with this specifically because selu *only* works on dense layers.

I chose Adam, even though this isn't a spare problem, because Adam excels on noisy problems and non-stationary objectives (definitely this), and because adam typically doesn't require a lot of hyperparameter tuning its ideal here, especially considering because I don't know what the hyperparameters should be to begin with.

I did work out some general guidelines on training quantity vs validation, etc.
The initial set wasn't huge or anything, roughly 110k pairs for training.

It converged pretty quick all things considered, and to the low loss like I mentioned, but even then the system always outputs the same result, regardless of the input, so obviously I'm doing something incorrectly.

The effectiveness of this approach for training and validation makes me question if I haven't got something wildly wrong. Still exploring though and figuring out how to get my answers back out. I'm hoping I just fucked up the output, and not the input as well.

Here's a story about why putting util functionality in a generic parent class is baaaad. So we run into a bug where an online shop module we develop causes a third party module to break the entire site until the session expires.

We track the bug down to the fact that the third party module has added some functionality to the part of the shop that deals with the cart and that functionality expects that one of the module's libraries is initialized. But as it turns out another of that module's libraries that is loaded earlier is fetching the cart and thus triggers our module which adds gifts to it.

Now, since we need a deeper integration with the cart to make gifts depend on the cart contents we call the part of it that now depends on the third party module's unloaded library.. So we think changing the order the third party libraries are loaded will fix the issue, only to discover the unloaded one is a child class of the first and the cart is fetched in the parent constructor. The parent of course then turns out to be a generic util class, inherited by all the module's libraries, so whatever order we load them in, the constructor is always called, so we had no other choice but to dynamically disable our module during the initialization of those libraries and then patch the updated cart contents into them after they've all been initialized.

At this point we get curious what that module's doing with the cart contents only to discover.. nothing. It's just that the parent class is full of utils and data fetching that the vendor reuses in all their modules..

Seriously, FUCK initialization in parenthesis.
Because of fucked-up C++ standards I spent half an hour debugging a cryptic AST error because my method definition was apparently parsed as a member initialization.

I had a pretty good year! I've gone from being a totally unknown passionate web dev to a respected full stack dev. This will be a bit lengthy rant...
Best:
- Got my first full time employment dev role at a company after being self-taught for 8+ years at the start of the year. Finally got someone to take the risk of hiring someone who's "untested" and only done small and odd jobs professionally. This kickstarted my career, super grateful for that!
- Started my own programming consulting company.
- Gained enough confidence to apply to other jobs, snatched a few consulting jobs, nailed the interviews even though I never practiced any leet code.
- Currently work as a 99% remote dev (only meet up in person during the initialization of some projects.) I never thought working remotely could actually work this well. I am able to stay productive and actually focus on the work instead of living up to the 9-5 standard. If I want to go for a walk to think I can do that, I can be as social and asocial as I want. I like to sleep in and work during the night with a cup of tea in the dark and it's not an issue! I really like the freedom and I feel like I've never been more productive.
- Ended up with very happy customers and now got a steady amount of jobs rolling in and contracts are being extended.
- I learned a lot, specialized in graph databases, no more db modelling hell. Loving it!
- Got a job where I can use my favorite tools and actually create something from scratch which includes a lot of different fields. I am really happy I can use all my skills and learn new things along the way, like data analysis, databricks, hadoop, data ingesting, centralised auth like promerium and centralised logging.
- I also learned how important softskills are, I've learned to understand my clients needs and how to both communicate both as a developer and an entrepeneur.
Worst:
- First job had a manager which just gave me the specifications solo project and didn't check in or meet me for 8 weeks with vague specifications. Turns out the manager was super biased on how to write code and wanted to micromanage every aspect while still being totally absent. They got mad that I had used AJAX for requests as that was a "waste of time".
- I learned the harsh reality of working as a contractor in the US from a foreign country. Worked on an "indefinite" contract, suddenly got a 2 day notification to sum up my work (not related to my performance) after being there for 7+ months.
- I really don't like the current industry standard when it comes to developing websites (I mostly work in node.js), I like working with static websites (with static website generators like what the Svelte.js driver) and use a REST API for dynamic content. When working on the backend there's a library for everything and I've wasted so many hours this year to fix bugs and create workarounds related to dependencies. You need to dive into a rabbit hole for every tool and do something which may work or break something later. I've had so many issues with CICD and deployment to the cloud. There's a library for everything but there's so many that it's impossible to learn about the edge cases of everything. Doesn't help that everything is abstracted away, which works 90% of the time but I use 15 times the time to debug things when a bug appears. I work against a black box which may or may not have an up to date documentation and it's so complex that it will require you to yell incantations from the F#$K
era and sacrifice a goat for it to work properly.
- Learned that a lot of companies call their complex services "microservices". Ah yes, the microservice with 20 endpoints which all do completely unrelated tasks?

[java] INFO [Initialization:101] FINISHED. The initialization took: 0d 00h:12m:00s:753ms
[java] INFO [Initialization:101] FINISHED. The updating took: 0d 00h:48m:01s:396ms
BUILD SUCCESSFUL
Total time: 72 minutes 51 seconds

... new (negative) record

... just fyi: no tests have been run, just a regular "hybris initialization" on monday morning ...

If you are reviewing pull requests to your code base, what aspects do you focus?
I am asking, because a colleague of mine most of the time complains about minor code style issues, like the following:

"why is that method static, it belongs to the object. I am not used to static methods, please adjust"

"the methods are not arranged in a standardized breadth-first order"

"Please use that other kind of initialization..."

Those nit-picks make me crazy sometimes. Is it just me?
I mean, her reviews also contain valuable content, but those small complains about coding style feel like "I did not found anything else but I have to". ^^

In my free time I'm developing a small init system with focus on parallel initialization.

!Rant

TL;DR - Getting married can lead to installing life malware.

MARRIAGE SOFTWARE....

A young husband wrote this to a Systems Analyst -
(Marriage Software Div);

Dear Systems Analyst,

I am desperate for some help! I recently upgraded my program from Girlfriend 7.0 to Wife 1.0 and found that the new program began unexpected Child Processing and also took up a lot of space and valuable resources. This wasn't mentioned in the product brochure.

In addition Wife 1.0 installs itself into all other programs and launches during systems initialization and then it monitors all other system activities.

Applications such as "Boys' Night out 2.5" and "Golf 5.3" no longer run, and crashes the system whenever selected.

Attempting to operate selected "Soccer 6.3" always fails and "Shopping 7.1" runs instead.

I cannot seem to keep Wife 1.0 in the background whilst attempting to run any of my favorite applications. Be it online or offline.
.
I am thinking of going back to "Girlfriend 7.0", but uninstall doesn't work on this program. Can you please help?

.... The Systems Analyst replied:

Dear Customer,
This is a very common problem resulting from a basic misunderstanding of the functions of the Wife 1.0 program.

Many customers upgrade from Girlfriend 7.0 to Wife 1.0 thinking that Wife 1.0 is merely a UTILITY AND ENTERTAINMENT PROGRAM.

Actually, Wife 1.0 is an OPERATING SYSTEM designed by its Creator to run everything on your current platform.
You are unlikely to be able to purge Wife 1.0 and still convert back to Girlfriend 7.0, as Wife 1.0 was not designed to do this and it is impossible to uninstall, delete or purge the program files from the System once it is installed.

Some people have tried to install Girlfriend 8.0 or Wife 2.0 but have ended up with even more problems. (See Manual under Alimony/Child Support and Solicitors' Fees).

Having Wife 1.0 installed, I recommend you keep it Installed and deal with the difficulties as best as you can.

When any faults or problems occur, whatever you think has caused them, you must run the.........
C:\ APOLOGIZE\ FORGIVE ME.EXE Program and avoid attempting to use the *Esc-Key for it will freeze the entire system.

It may be necessary to run C:\ APOLOGIZE\ FORGIVE ME.EXE a number of times, and eventually hope that the operating system will return to normal.

Wife 1.0, although a very high maintenance programme, can be very rewarding.

To get the most out of it, consider buying additional Software such as "Flowers 2.0" and "Chocolates 5.0" or "HUGS\ KISSES 6.0" or "TENDERNESS\ UNDERSTANDING 10.0" or "even Eating Out Without the Kids 7.2.1" (if Child processing has already started).

DO NOT under any circumstances install "Secretary 2.1" (Short Skirt Version) or "One Nightstand 3.2" (Any Mood Version), as this is not a supported Application for Wife 1.0 and the system will almost certainly CRASH.

BEST WISHES!

Yours,"
Systems Analyst.
-----------------------------------------------
I'm not sure if this is a repost - if it is I apologise, but it's too good not to share.

It's high time people start understanding the difference between 0 (zero) and NULL.

Zero: You visiting the toilet and notice the exhausted toilet paper roll.

NULL: You visiting the toilet and notice that there's no toilet paper roll.

Get the difference?

1) Use Jmockit to write some test utilizing $clinit method

2) Have private static field in enum (along with static initialization block)

????(Doesn't work when I run mvn clean install (despite the fact it was working on eclipse) on cmd, because classes can't be found, log4j, etc.)

Profit

#Suphle Rant 11: Laravel board launch

The launch took almost 2 weeks more than originally slated, because I sought to install it manually, just as an outsider would. Installation steps had been documented, automated tests for the installation tests were passing. When time came to actually execute the binary from the terminal, we went from one obstacle to the other. First, were the relatively minor Composer/Roadrunner issues, eventually resolved by the helpful RR maintainers who sat with me through a Discord server for about 2 hours until their command ran the way I needed it to.

Next was the Psalm scare: One of my value propositions was the guarantee of eliminating all type related bugs in Suphle apps. I intended to use Psalm for that. Wrote tests as usual. Turns out the library behaves differently under conditions differing from raw CLI usage. I resurrected threads I'd opened since December that were left unattended, and with some help from the maintainer, we eventually got it to do what I need it to do.

I was all the more frightened by the fact that Transphporm had caused me to renege on one of my earlier promises. I can only miss so many targets. After this, the docs had to be updated with all the changes effected to accurately integrate those two. Project installation and initialization commands were ran rigorously to ensure all progresses smoothly.

Tagged one final release and suddenly became impatient to launch on our local Laravel group chat where I've been a member for the last 4+ years, where we've had a rollercoaster of emotions. In that time, I've refined my launch speech to suit that audience -- obviously, countless times. Not just a tame "It's my pleasure to announce what I've been working on", but near 40 messages going into details about the inner workings, why it was built, how it compares. An expose that dove deeper than I would anywhere else.

I scheduled a time for them to tune in and got some encouraging anticipation. Ended up deflated after posting the whole thing. Only about 5 persons interacted. 1 (who I've chatted with outside the board) was quite enthusiastic. Feverishly checked the docs but commented it was overwhelming and he'd need more time. Already starred the repository.

For some context, there are give or take 250 members on that board. Not all are active but activity there easily reaches a crescendo when the topic discussed is about inanities like what 3rd party services to use for SMS, how to receive salaries from abroad, or job openings. I was optimistic when the acquaintance mentioned above published a payment library and met a riotuous welcome as one of their own. Maybe, they are simply not fond of me and the speech should have been passed off to someone else.

I checked Packagist installs -- not more 10. For 3 years, I'd been hyped up for that night; but for some reason, the audience I considered myself closest to flopped, woefully. Thankfully, this isn't the main launch. I'm still holding out hope for that. If it fails, I would have sunk an immeasurable amount of effort and time, that nobody will compensate me for. That is the one place I go to see those more advanced than me in PHP. I constantly learn there and find stimulating conversations there.

Now, I can no longer predict reception from other presentations. All I can do now is hope

Is relying on probability bad practice? I have a container that needs to know about all other instances of the same container and assign a unique ID to itself on initialization. I thought that if I selected a big random number as its id on startup, the chances of it having the same id as another instance is very low.

To prevent two instances from having the same id I could check for all running instances on the network, but what if two instances start at the same time? They won't find each other since none of them will be fully initialized when their id checks run.

Probability says I'll be fine, Murphy's law says I won't. What do?

Commons sense/ best practice:

Is It ok ti initialize (angular) variabile as {id:" ", name:" ", ..} to avoid errors in the browser console such as "can't get ID of undefined"?

My concern is code readability and debugging, is not ok for the ones looking at the browser console to have such useless errors, on the other side you have to initialize some variables with object that have a lot of keys(id name ecc...) Whith empty fields...useless.

The apps work both cases, whit or whitouth initialization.
By the way we are getting such data by api calls later on.

Just finished my finals.
Had to run k-means with pen and paper only. I find this kind of question stupid but why not. BUT WHY THE FUCK DID YOU CHOSE SOME INITIALIZATION THAT TAKES 13(!!!!) FUCKING ITERATIONS TO CONVERGE ? Just in case my first 12 iterations are correct by chance ? Guess what, you fucktard, I GOT IT.
And doing the same calculus by hand 13 FUCKING TIMES is moronic as hell, you retarded piece of shit ! When you train your neural networks, do you also backpropagate your gradient all by yourself, mongoloid baboon? Getting sick of those stupid assignements

What is the Hot Tech (Best Emerging Technology) of this generation?
Machine Learning, Internet of Things, Big Data, Android development.

The build broke right before code review with the Lead. It didn't just break in one place, oh no, that would be too simple. It broke everywhere, right down to the core mechanic. I spent the next 3 hours trying to find out why it broke, checking everything involved in this part of the system. It was a freaking Initialization call placed inside a conditional statement instead of outside.

#Suphle Rant 6: Deptrac, phparkitect

This entry isn't necessarily a rant but a tale of victory. I'm no more as sad as I used to be. I don't work as hard as I used to, so lesser challenges to frustrate my life. On top of that, I'm not bitter about the pace of progress. I'm at a state of contentment regarding Suphle's release

An opportunity to gain publicity presented itself last month when cfp for a php event was announced last month. I submitted and reviewed a post introducing suphle to the community. In the post, I assured readers that I won't be changing anything soon ie the apis are cast in stone. Then php 7.4 officially "went out of circulation". It hit me that even though the code supports php 8 on paper, it's kind of a red herring that decorators don't use php 8 attributes. So I doubled down, suspending documentation.

The container won't support union and intersection types cuz I dislike the ambiguity. Enums can't be hydrated. So I refactored implementation and usages of decorators from interfaces to native attributes. Tried automating typing for all class properties but psalm is using docblocks instead of native typing. So I disabled it and am doing it by hand whenever something takes me to an unfixed class (difficulty: 1). But the good news is, we are php 8 compliant as anybody can ask for!

I decided to ride that wave and implement other things that have been bothering me:
1) 2 commands for automating project setup for collaborators and user facing developers (CHECK)
2) transferring some operations from runtime to compile/build TIME (CHECK)
3) re-attempt implementing container scopes

I tried automating Deptrac usage ie adding the newly created module to the list of regulated architectural layers but their config is in yaml, so I moved to phparkitect which uses php to set the rules. I still can't find a library for programmatically updating php filed/classes but this is more dynamic for me than yaml. I set out to implement their library, turns out the entire logic is dumped into the command class, so I can neither control it without the cli or automate tests to it. I take the command apart, connect it to suphle and run. Guess what, it detects class parents as violations to the rule. Wtflyingfuck?!

As if that's not bad enough, roadrunner (that old biatch!) server setup doesn't fail if an initialization script fails. If initialization script is moved to the application code itself, server setup crumbles and takes the your initialization stuff down with it. I ping the maintainer, rustacian (god bless his soul), who informs me point blank that what I'm trying to do is not possible. Fuck it. I have to write a wrapper command for sequentially starting the server (or not starting if initialization operations don't all succeed).

Legitimate case to reinvent the wheel. I restored my deleted decorators that did dependency sanitation for me at runtime. The remaining piece of the puzzle was a recursive film iterator to feed the decorators. I checked my file system reader for clues on how to implement one and boom! The one I'd written for two other features was compatible. All I had to do was refactor decorators into dependency rules, give them fancy interfaces for customising and filtering what classes each rule should actually evaluate. In a night's work (if you're discrediting how long writing the original sanitization decorators and directory iterator), I coupled the Deptrac/phparkitect library of my dreams. This is one of the those few times I feel like a supreme deity

Hope I can eat better and get some sleep. This meme is me after getting bounced by those three library rejections