To become an engineer (CS/IT) in India, you have to study:
1. 3 papers in Physics (2 mechanics, 1 optics)
2. 1 paper in Chemistry
3. 2 papers in English (1 grammar, 1 professional communication). Sometimes 3 papers will be there.
4. 6 papers in Mathematics (sequences, series, linear algebra, complex numbers and related stuff, vectors and 3D geometry, differential calculus, integral calculus, maxima/minima, differential equations, descrete mathematics)
5. 1 paper in Economics
6. 1 paper in Business Management
7. 1 paper in Engineering Drawing (drawing random nuts and bolts, locus of point etc)
8. 1 paper in Electronics
9. 1 paper in Mechanical Workshop (sheet metal, wooden work, moulding, metal casting, fitting, lathe machine, milling machine, various drills)
And when you jump in real life scenario, you encounter source/revision/version control, profilers, build server, automated build toolchains, scripts, refactoring, debugging, optimizations etc. As a matter of fact none of these are touched in the course.
Sure, they teach you a large set of algorithms, but they don't tell you when to prefer insertion sort over quick sort, quick sort over merge sort etc. They teach you Las Vegas and Monte Carlo algorithms, but they don't tell you that the randomizer in question should pass Die Hard test (and then you wonder why algorithm is not working as expected). They teach compiler theory, but you cannot write a simple parser after passing the course. They taught you multicore architecture and multicore programming, but you don't know how to detect and fix a race condition. You passed entire engineering course with flying colors, and yet you don't know ABC of debugging (I wish you encounter some notorious heisenbug really soon). They taught 2-3 programming languages, and yet you cannot explain simple variable declaration.
And then, they say that you should have knowledge of multiple fields. Oh well! you don't have any damn idea about your major, and now you are talking about knowledge in multiple fields?
What is the point of such education?
PS: I am tired of interviewing shitty candidates with flying colours in their marksheets. Go kids, learn some real stuff first, and then talk some random bullshit.

Summary: Burnout, and everything's broken.

I don't feel like doing a damn thing today. I look at the code and cringe. I look at Slack and think "ugh. i can't." Mental capitals are even too much work.

(I've started reading "Zen and the Art of Motorcycle Maintenance" to try and combat burnout. I'll write a rant/story about it here if I find it helpful. but all I want to do today is drink tea and read.)

But onto the story:

Heroku is deprecating support for and will automatically upgrade any old verisons of Postgres running on its platform after August something (like five days from now).

I performed the upgrade to PG10 on Sunday (and late into the night), provisioning a new follower, blah blah blah.

However, the version of Rails we're using (4.2.x) doesn't support PG10 sequences, so I manually added in support via a monkeypatch. I did this on our QA servers first, obviously, and everything worked as expected. After half a day of no issues, I did the same on production, and again: everything worked as expected.

But today? I keep hearing about new things that are broken. One specific type of alert doesn't work for one specific person (wat). Can't send [redacted] at all. Can't update merchants! Yet there are magically no errors logged.

That last one (well, two) are just great; let me explain: when there's an error concerning merchants, the error gets caught, isn't logged or recorded anywhere so it just disappears, and the rescue block triggers a json response instead and happily exits. This is for an internal admin tool, so returning a user-friendly error is kinda stupid anyway, but masking what actually happened? fuck that dev with an obelisk made from spikes and solidified pain. That json response is also lovely: it's a 200 OK returning {status: 1, data: "[generic message containing incorrect IT jargon]"}. Doesn't even say "error" anywhere. Bloody everything about this pattern is absolutely wrong. Even the friggin' text.

Fucking hell. I want to pipe the entire codebase into shred and walk out the door.

But I digress. So many things are broken, my motivation is wanning to a sliver, and I have a conference call today where I'll undoubtedly be asked why everything is on smoking and/or on fire, and my huge and overly productive week last week will ofc mean nothing by contrast.

Ugh.

`shred ~/dev/work -zfu -n 32 &; ./brew tea --hot && wine ~/takeabreak.exe`

I've assembled enough computing power from the trash. Now I can start to build my own personal 'cloud'. Fuck I hate that word.
But I have a bunch of i7s, and i5s on hand, in towers. Next is just to network them, and setup some software to receive commands.

So far I've looked at Ray, and Dispy for distributed computation. If theres others that any of you are aware of, let me know. If you're familiar with any of these and know which one is the easier approach to get started with, I'd appreciate your input.

The goal is to get all these machines up and running, a cloud thats as dirt cheap as possible, and then train it on sequence prediction of the hidden variables derived from semiprimes. Right now the set is unretrievable, but theres a lot of heavily correlated known variables and so I'm hoping the network can derive better and more accurate insights than I can in a pinch.
Because any given semiprime has numerous (hundreds of known) identities which immediately yield both of its factors if say a certain constant or quotient is known (it isn't), knowing any *one* of them and the correct input, is equivalent to knowing the factors of p.

So I can set each machine to train and attempt to predict the unknown sequence for each particular identity.

Once the machines are setup and I've figured out which distributed library to use, the next step is to setup Keras, andtrain the model using say, all the semiprimes under one to ten million.

I'm also working on a new way of measuring information: autoregressive entropy. The idea is that the prevalence of small numbers when searching for patterns in sequences is largely ephemeral (theres no long term pattern) and AE allows us to put a number on the density of these patterns in a partial sequence, but its only an idea at the moment and I'm not sure what use it has.

Heres hoping the sequence prediction approach works.

!(!(!(!(!(!(!(!rant)))))))
My new HTC smartphone hates me.
First it started to shut down all of the sudden yesterday night, when I was solving quadratic equations on my laptop.
I thought that it might be due to low battery. So I have restarted it. After putting itself into a bootloop for 4 start sequences, it was able to fully start to the page where it told me to enter the security pin to decrypt my files. I also had 30 attempts left. Like a ransomware.

I was like "tf I didn't set anything up".
So I decided to use my first attempt as I had 30 attempts left.
I entered the pin (I can swear that it's correct) and it told me that it has to wipe the /data partition.

I did that. I pressed that button. After waiting for 30 minutes I gave up and rebooted into the bootloader.
Bootloader -> Download Mode -> wipe /data (stock rom + stock recovery btw.)

Some error with "e: mount /cache failed[...]e: mount /data failed"
So, I tried using the adb sideload - no success.
Fastbooted into RUU Mode - HTC keeps rebooting itself into the RUU Mode - no success

Tried to flash the firmware and twrp recovery from Download mode - no success
Then I tried to flash all these things from the sd card - no success

Searched for revolutionary (I know this from my old HTC sensation device).
It wasn't big of any help.

Then someone on xda recommended htcDev (htc's <b>developer-friendly</b> lol site)
I followed every step. Everything seemed to be okay.
I got to the last step.
I needed to get my encrypted token by entering "fastboot oem get_identifier_token" to be able to submit it to HTC, and after they would send me an e-Mail with an .bin file that would let me unlock the bootloader to be able to flash my way through all this headache giving fucking piece of dog shit!

But since I can't back to the phone settings to select the bootloader activation box that would let me get my token... but nah.

FML

------------
Sent by using the devRant web app (:\)

Not work, but was very pissed off anyways.

So, today my C# lecturer was teaching about escape sequences in strings. Specifically, he's showing how to escape the single quotes character ( ' ) since we're learning about how to send SQL queries as well.
He started writing on the whiteboard the following and said that this was how to escape the single quotes character in a string:
\' "abc123" \'
Me and one of my classmates looked at this and started to ask questions, since this is definitely not how you do it. Somehow, the lecturer could not understand us. We tried to explain it the best we could, starting from verbally, then writing on the whiteboard, then even showing code on a laptop. For some unknown reason the lecturer still couldn't understand where he was wrong and both of us just gave up after 15 minutes of trying to explain it.
Mind you, most of the class had little to none prior programming experience, me and said classmate are one of the few that actually programmed before, so all my other classmates were just very confused as to what is right and what is wrong.
Now I'm really questioning my lecturer's abilities....

Every time I think I've seen the worst there's someone to prove me wrong...

GRANT ALL ON ALL TABLES/SEQUENCES
To web user on production

if (x == 1) y = 1;
else y = x;

loop through a collection and get 'few' relations using ORM - 1000s of queries and not a single join - but don't worry, "The ITs will just add more RAM and some CPU cores to the server"

4th day off and I already miss this

One thing when working with a ton of data:
If there is a slight, infinitesimal probability that something will be wrong, then it will 100% be wrong.

Never make assumptions that data is consistent, when dealing with tens of gigabytes of it, unless you get it sanitized from somewhere.

I've already seen it all:
* Duplicates where I've been assured "these are unique"
* In text fields that contain exclusively numeric values, there will always be some non-numeric values as well
* There will be negative numbers in "number sequences starting with 1"
* There will be dates in the future, and in the far far future, like 20115 in the future.
* Even if you have 200k customers, there will be a customer ID that will cause an integer overflow.

Don't trust anything. Always check and question everything.

Can someone please tell GitHub to add Vim mode? I'm tired of filling my comments with Vim navigation sequences only to have to correct them with a mouse...

😫

While I was exploring multiplication tables I stumbled on something cool.
Take any power of 2 on the multiplication chart.
Now look at the number in the bottom left adjacent box.
The difference of these two numbers will always be a Mersenne number.
Go ahead. Starting on the 2's column of a multiplication table, look in the bottom left of each power of 2 and get the difference.

2-2 = 0
4-3= 1
8-5 = 3
16-9=7
32-17=15

etc.

While the online journal of integer sequences lists a lot of forumlas, I just wrote what came to mind (I'm sure its already known):
((2**i)-(((2**i)/2)+1))

The interesting thing about this is it generates not only the Mersenne numbers, but if you run i *backwards* it generates *additional* numbers.
So its a superset of mersenne numbers.

at i = 0 we get -0.5
i=-1 -> -0.75
i=-2 -> -0.875
i=-3 -> -0.9375
i=-4 -> -0.96875

And while this sequence is *not* mersenne numbers, mersenne numbers *are* in this set.

Just a curious discovery is all.

A gene development framework that could:
* simulates the resulting proteins correctly (folding and all)
* debug genetic sequences/proteins
* Design injectors for gene sequences into an existing genetic sequence
Etc'

Client be like:
Pls, could you give the new Postgres user the same perms as this one other user?

Me:
Uh... Sure.

Then I find out that, for whatever reason, all of their user accounts have disabled inheritance... So, wtf.

Postgres doesn't really allow you to *copy* perms of a role A to role B. You can only grant role A to role B, but for the perms of A to carry over, B has to have inheritance allowed... Which... It doesn't.

So... After a bit of manual GRANT bla ON DATABASE foo TO user, I ping back that it is done and breath a sigh of relief.

Oooooonly... They ping back like -- Could you also copy the perms of A on all the existing objects in the schema to B???

Ugh. More work. Lets see... List all permissions in a schema and... Holy shit! That's thousands of tables and sequences, how tf am I ever gonna copy over all that???

Maybe I could... Disable the pager of psql, and pipe the list into a file, parse it by the magic of regex... And somehow generate a fuckload of GRANT statements? Uuuugh, but that'd kill so much time. Not to mention I'd need to find out what the individual permission letters in the output mean... And... Ugh, ye, no, too much work. Lets see if SO knows a solution!

And, surprise surprise, it did! The easiest, simplest to understand way, was to make a schema-only dump of the database, grep it for user A, substitute their name with B, and then input it back.

What I didn't expect is for the resulting filtered and altered grant list to be over 6800 LINES LONG. WHAT THE FUCK.

...And, shortly after I apply the insane number of grants... I get another ping. Turns out the customer's already figured out a way to grant all the necessary perms themselves, and I... No longer have to do anything :|

Joy. Utter, indescribable joy.

Is there any actual security reason for disabling inheritance in Postgres? (14.x) I'd think that if an account got compromised, it doesn't matter if it has the perms inherited or not, cuz you can just SET ROLE yourself to the granted role with the actual perms and go ham...

I wonder if anyone has considered building a large language model, trained on consuming and generating token sequences that are themselves the actual weights or matrix values of other large language models?

Run Lora to tune it to find and generate plausible subgraphs for specific tasks (an optimal search for weights that are most likely to be initialized by chance to ideal values, i.e. the winning lottery ticket hypothesis).

The entire thing could even be used to prune existing LLM weights, in a generative-adversarial model.

Shit, theres enough embedding and weight data to train a Meta-LLM from scratch at this point.
The sum total of trillions of parameter in models floating around the internet to be used as training data.

If the models and weights are designed to predict the next token, there shouldn't be anything to prevent another model trained on this sort of distribution, from generating new plausible models.

You could even do task-prompt-to-model-task embeddings by training on the weights for task specific models, do vector searches to mix models, etc, and generate *new* models,
not new new text, not new imagery, but new *models*.

It'd be a model for training/inferring/optimizing/generating other models.

I keep obsessing about the time. I keep looking for patterns as the day goes on. I noticed the other day that these patterns for time have math relationships in each:
15:18:21 ie 15 + 3 = 18 and 18 + 3 = 21
03:18:21 ie 3 + 18 = 21

One time is morning at 3 am the other is evening at 3 pm. I cannot stop looking for these.

Is the descent into madness?

Trees -> declarative programming

Loops -> functional programming

Sequences -> imperative programming

Graphs -> dynamic programming

Good mapping, yeah or no?

headaches..

Likely a very simple algorithm to get these packets processed in the right order, got plenty of metadata that should help, but since yesterday I am wrapping my head around it and cannot figure out a sane way...

Build a automatic diagnostic system that sequences your DNA and runs blood tests in hours, then creates personalised medication for your illness.

I've gotten started with web dev in the past and learned HTML and CSS and started learning JS but I never could understand what I could use for a code editor to practice and pretty much forgot all of that stuff. Now I'm trying to learn Python, but what's pissing me off is paying for a phone app that doesn't teach you to write code in these lessons, rather interactive multiple choice questions and "put this in the right order". sequences. This is not learning for me, this is informing. Which is info I don't retain. And If i'm paying for it why is there so little to these lessons? Barely covering anything. I've done every lesson Mimo had for python but it didn't really explain the practicality of what it was teaching me and they skipped a lot of shit. Changing the pace of the lesson from Print this and that and heavily explain the most basic stuff 3x over to only explaining the more advanced stuff one fucking time.

I would really like learning python while being walked through a project as a lesson. Teach the terminology, structure, application, process, rinse and repeat, and outcome all in one. With a project target to look forward to. I need a goal to keep my interest.

So far all I know about python is its a programming language used to create Youtube. And I'm trying to learn it because I keep reading that its the recommended starting line. But I need to be able to visualize what this code can be used for. Explanations in terminology I haven't been taught yet just frustrates me. And I read everyone's posts and see many people mention being frustrated, but I haven't even started coding yet. Feel free to comment and redirect me to page that can help. Links are appreciated. Nay, encouraged!

For years now I've been "dreaming in code" but in the stupid way, which is only appropiate.

I try to explain it to myself and *I* can't understand it.

One, by some oniric enchantment, is capable of communicating signals through use of some symbolic language; and any time one speaks, they are affecting all that follows.

So a sequence of these, of any size, corresponds to some kind of program, and the self is some sort of collection of mutable structures being affected by them. And new symbols arise from within the self, corresponding to sequences of previously spoken symbols.

This process in itself can be satisfying, for the mere challenge of engaging with it's bottomless complexity, but it also suffers from a complete lack of purpose.

What does it mean? It's all undefined, yet doing something, so it must *mean* something. But what is it doing? One simply cannot grasp it!

I go to bed at night and traverse my tree, I recognize it, I've been working on it for years. Time is different there, you can just keep infinitely building shit, it never ends. Then I wake up and everything makes sense, for a little while.

But what I see isn't quantifiable; I can't turn it into a representation that works outside of a dream. Does it give me some vague ideas for the "actual" code I'm working on, yes of course. Yet it's all so... elusive, I can never put it into words. How exactly I could think of this? Well, it's in my tree, I know it because I wrote it as I slept. But how?

Fucking brains, maan.