cue_the_strings

Location: Ljubljana, Slovenia

Remote: yes

Willing to relocate: No

Technologies: C++, Qt, C, Python (Jinja2, Numpy, SciPy, SymPy, PyQt5, Flask, Sqlalchemy), CMake, Make, CI, unit testing (Google test, Unity), static analysis (writing tools with Clang tooling), Ansible, Docker / Podman, POSIX shell, Bash, Linux administration, Linux packaging, LaTeX, ReST; Also familiar with HTML, CSS, ES6+, React, Node.js, but not an expert; Dabbled in Rust, willing to learn it.

Résumé/CV: on request; CS BSc, 6 years of experience

Email: andrej.radovic@protonmail.com

---

I'm looking for a remote C++ development job: backend/libraries/GUI all suits me fine, just not embedded. Dealbreakers: "defense" (weapons/war), "gaming" (gambling) and crypto.

Up until now I did embedded development (in spite of my CS background), along with Qt C++ development (6 years of experience). Now I want to switch to (non-embedded) C++ full time. I can say I'm very proficient at modern C++, I've devoted most of my free time for 2 years to doing a deep dive of it.

I'm also very experienced with Python: generating C/C++ code, writing development tools with it, including Flask web-based tools and webhook-glue-services, as well as processing numerical data and doing some DSP in it.

I'm an expert at setting up company infrastructure (with Podman in Ansible on RHEL), administering and connecting internal services (like Gitea, Gitlab, Sonarqube, Reviewboard) and setting up CI systems (Drone, Gitlab CI) for C++/C/Python development. I'm also very good at large-scale refactoring, writing static analyzers and fixers with Clang tooling (clang-tidy). Automation of all kinds is my game.

I have several small FOSS projects on my Gitlab (https://gitlab.com/andrejr), packages on PyPI (https://pypi.org/user/andrejr/) and I also maintain packages on AUR (https://aur.archlinux.org/account/andrejr).

I'm willing and able to establish, recruit and lead a small local team in Slovenia.

Yes, I've been considering that since the beginning. Website (backend in C++, frontend in Svelte) takes priority, because this solution is good enough so far, and I'd really like to have access to my ZK on my phone. Probably not a website meant for the open internet: I have a server at home + use Wireguard VPN, so my phone can connect to local services/sites at all times.

I decided on the FOSS option even before GitHub sold out to Microsoft. I've used Gitlab (and Gitlab CI) on my home server for years, it's been excellent so far.

Tbh I use it rather differently, mostly for programming stuff. It makes sense to have the following things connected, for example: visitor pattern, std::visit, overload lambda template. Makes it easy to find related concepts and tricks. All bite sized snippets and explanations.

It's also great for storing rarely used commandline snippets and tricks in a structured way, for things I use once in several years. Picocom and screen tricks and configs neatly connected to pages for old lab and network equipment, MIDI sysex tools and tricks connected to docs about certain FX processors and synths,...

It's especially useful when you simply forget that you even had solutions for certain things, and that certain things are connected. I did so many vastly different things over the years that this seems to be the only way to efficiently jump back into something I haven't done in a while. I have a huge personal wiki (2000+ articles in vimwiki) which I'm now migrating to this tool, and there's one huge difference: I'm never sitting there thinking "How do I jam this new article into the wiki hierarchy... Oh, it's equally important for both of these categories...". I just jot it down in vim, add a couple tags, maybe a link or two, maybe find similar articles for inspiration if I don't know what to link to, and done. Extremely fast compared to a big vimwiki.

I've developed a set of UNIX-philosophy CLI utilities for managing your zettelkasten, a C++ library (used by the utilities) and a set of shell scripts that makes the whole thing easy to use for the terminal-savvy. The fzf-based finder (script) integrates with Vim. It's probably the fastest set of zettelkasten utilities. Heavily inspired by sirupsen/zk, but faster, more powerful and more consistent interface.

utilities + lib: https://gitlab.com/andrejr/zetk shell wrappers: https://gitlab.com/andrejr/zetk_fzf

Shared under AGPL3.

As a recovering embedded engineer (though I am a CS grad), I strongly suggest noone enter the field: the pay is miserable, the job market is saturated with EE grads who wouldn't be able to do regular development without a major time investment.

I'm switching to full-time C++ backend and GUI development (so far it was 70% embedded and 30% this. I find that there is much more complexity in developing server-client systems, which optionally integrate with other systems, than it is to write low level code (maybe I'm just good at writing low level code). It's more intelectually rewarding, and you get to use higher level concepts. There's always room for advancement and computer programming is deep in a way that embedded engineering never will be. I've worked on embedded systems for 6 years and I've tried pretty much everything there is to try, from the lowest level to graphics from scratch, RTOS development and networking.

This is something that's completely obvious to me. What I don't understand is why people online insist that Rust is so different and C++ knowledge is useless - I doubt I'd grok it immediately if I didn't understand C++.

I'd say C++ at this point in time. It's getting better by the day, yet it has decades of high quality projects and libs you can rely on. Each new version is better than the last. It may be irritating that few things are ever hard-deprecated, and you have to keep a list of current best practices. On the other hand, concepts, ranges, previously auto and constexpr, are all godsend and make for an expressive language that's still very powerful.

With Rust, depending on what you're working on, you may lose almost no time integrating C and C++ (with an extern C interface) libraries, or you may lose 50% of it. Sure, you don't have to know CMake at least (which is a hard requirement for C++ nowadays), but using any extraordinary library dependency in Rust may prove to be a nightmare. Essentially: if someone else didn't already package it, prepare to lose some serious time. The language is nice, but the ecosystem is immature.

Things are probably going to change, and Rust will become more viable over time, but C++ is also improving at a rapid pace so I guess Rust will never completely outcompete it.

I'm an avid music listener (in my 30s), and it'd be hard for me to name just 1 (or 10 or even 30) of my favorite songs. A couple of albums I keep in high regard came out in the last several years. Also, I seem to wear out music over time: you can only listen to Loveless or SAW 89-92 or Astral Weeks or Velvet Underground and Nico so many times. I know, yes, those albums are some of the best music made, but I don't find pleasure in re-listening them for the 100th time - I'd rather listen to something new.

I probably took after my dad when it comes to music listening. He's in his 60s and still actively searching for and finding new favorites.

I knew a guy who swapped his World of Warcraft addiction with heroin. It worked surprisingly well, he never played WoW again. Last time I saw him, he tried 'borrowing' some money from me for 'medicine for his mom'.

CentOS Stream's last podman major version update broke 70% of my company's infrastructure, affecting 120+ employees and our customers. It was a bug that couldn't be fixed without reverting the version, but we had to revert the whole update because all kinds of dependencies also got updated. We switched to Rocky the next day, on all VMs, and haven't had trouble since. We never had trouble with the original Centos (7 and 8).

Hacking in some form (essentially being a script kiddie) isn't that difficult for a kid, especially if there is something to gain - bragging rights at least.

We got internet when I was 9. By the time I was 11 in the early 2000s, I was a full-blown script kiddie, scouring astalavista.box.sk for tools and scripts to hack anything that I could. I'd also hang on IRC. I thought hacking was the coolest thing ever, and took any opportunity to access anything forbidden or leave a silly signature anywhere. I didn't discriminate. It didn't matter that I had zero connection with any of these websites and services I 'hacked', I thought I was so cool.

Most of the stuff I did was simply following the instructions / tutorials, mostly Google dorking for strings in php files and using documented exploits. I never succeeded in any sort of targeted attack, and I lived in such a shithole that nothing important was online at the time, anyway. All of this was extremely low impact.

The only thing even remotely impressive about any of this was that I initially had very poor control of English. I was using a primitive, non-phraseological dictionary all the time, and somehow understood enough to actually apply some of the tutorials I read. People don't give kids enough credit for their determination and persistence, and kids have so much of free time to learn whatever they set their mind to.

There's open source, and then there's open source.

vscode, .net core, redis, mysql, qt vs vim, dwm (suckless jn general), weechat, ...

You should finish that thought: and the manufacturers won't invest their money to certify something that's open source and can be used by everyone, including their competitors.

If there was a will to do something about it, GPL3 wouldn't be in the way.

You're probably using Firefox anti-tracking capabilities or something similar, that's what caused issues for me on both Paypal and Amazon.

> Mercedes A series parts are affordable > The headlight is only sold together with an electronics module for steerable headlight control attached, for a cool 800€.

That module is positioned in such a way that water collects around it if the car gets sufficiently splashed from the front, and it dies even though the PCB is conformally coated.

I chuckled... sensibly.

I'd summarize the look as: decent quality on the faces, some parts of the skin randomly smoothed (like an amateurish blemish removal filter), hair always like it was taken with a 2015 mid range cell phone with HDR (say Motorola Moto Style).

I tried a blind test some time ago, 5 generated, 5 real faces, and was 100% correct.

How? The generated faces have too perfect teetch, sometimes illogical changes in skin texture and hairiness,... But one thing really stands out on its own and I think I could be just as correct if I only relied on it.

See how all these pictures are propely lit and exposed and seem like they were taken with a decent camera? Now notice how this isn't the case with hair, especially transparent parts of loose long hair, where you see individual hairs. These parts look like they have way more noise and like they've been taken in low light with high ISO, somewhat like in a bad HDR composite. This may not work for slicked hair or bald people, yet it's pretty evident in almost all other cases.

Stability on macOS? Laughable.

Macs are pretty popular among musicians and producers. When El Capitan dropped, pretty much all audio hardware and software stopped working. I remember that you couldn't use Native Instruments plugs and the majority of pro audio interfaces stopped working.

This was fixed after a while, but it took at least half a year for everything to work again for 50% of users and a full year for everything to work for 90% of users.

If you upgraded by accident and didn't have backups, you were in trouble. Even if you had backups, you could still have issues.

Here's a link from back in the day:

https://macprovideo.com/article/audio-software/5-reasons-mus...

Now, did something similar ever happen on Linux or Windows?

I can safely say that a lot of people lost some revenue because of that upgrade (there were also issues for video production software), at least until the word spread that you shouldn't upgrade.

True. On the other hand, thousands of people are dying in hospices and care homes, Covid or not.

Mortality increase (compared to the average of last n years) seems like the only good measure of Covid impact.

I first heard about this as a kid in 1999. I didn't even have internet back then (only got it in year 2000), and my country had just been bombed by NATO (your regular bombing campaign sprinkled with bombing civilians, as is tradition) and was lead by Slobodan fucking Milošević (a colossal piece of shit), so all kinds of conspiracies were flying around, everything was disinformation, and I didn't put much weight on this story even though my parents discussed it with friends several times during the next decade, which is why I remember it.

My parents' friend who had lived in New Zealand for a decade at that point came to the country to bury his father and was over at our house for dinner. He's an electrical engineer and used to work for a telco in New Zealand. He had heard about this ECHELON thing from multiple colleagues and found it plausible enough to talk about. He was told that all internet and phone traffic was being recorded realtime. Mind you, this was pre-9/11, so no 'terrorist threats'. The world ran on shitty computers and expensive storage and voice in any form (even compressed telephony) was considered bulky at the time. And this is 15 years prior to the machine learning boom, voice transcription was really bad, I still doubt it was used at that point, even though he mentioned it.

But he had witnessed an expansion of fiber and datacenter capacity sometime in 1997-99 and it seemed plausible to him (a real serious and professional guy).

Imagine my surprise when it turned out to be true. I mean, the story slowly became less and less far fetched over the following decade, but was still a total surprise when Snowden blew the whistle.

Powershell, for example, tries to address these points by naming commands more verbosely, having verbose flags and even having programs input and output object-y data instead of just plain text.

These are all nice ideas that don't work very well in practice because the verbosity is only beneficial when writing scripts and is just cumbersome for interactive use. Also, Linux (and other Unix-likes in general) are better designed to be used w/ scripted and cli tools, and the object nature of IO often just gets in the way. But the biggest issue is that the built-in tools just arent as powerful as POSIX core utilities and especially GNU-extended ones.

ITT: nobody takes readability into consideration + default Windows, Mac, Office fonts

Times has some objective merits, bland and ugly as it is. It is somewhat readable on screen due to hinting. It's available in Office and used to be the default, and many people won't even look for other Office fonts. If they did, they'd find that Georgia rules the 'works on any resolution + paper' category. Another (non-Office) solution is FOSS favorite Bitstream Charter (I suggest a variant called XCharter for better Unicode coverage). Both Georgia and Charter were designed by Matthew Carter.

If you have a high-dpi screen and a 1200dpi printer, you may as well use Garamond. Minion (or FOSS-licenced Cochineal) are a bit plainer. Linux Libertine is quite nice. If I wrote in a WYSIWYG editor, I'd probably still use Charter, because it's even simpler and cleaner, and print something else. These 'nice' serifs may provide some inspiration with their sheer beauty, but I find looking at them on a screen tiring and prefer the simplicity of Charter.

When it comes to sans-serifs, fortunately Arial isn't on the list. Both Arial and Helvetica are shitty choices for copy, Univers being just slightly better (wider spaced and easier to read) but still shit. Even Calibri reads significantly easier than all of them neo-grotesques.

Frutiger (and its near-clone Myriad) reads much easier on screen and in print. Syntax looks stylish on paper but crappy on screen, and same can be said of FF Meta (I blame the angled terminals on some letters). Fira Sans (Meta derivate) looks great on screen and on paper. If i had to use a sans-serif in a WYSIWYG editor, I'd pick Fira or Frutiger.

However, since I only write LaTeX and (Pandoc) Markdown (both in Vim), I write in Terminus or Inconsolata LGC, depending on screen resolution.