SadWebDeveloper

On the free and "AI Plus" plan you are only allowed 3.6, 3.1 Pro and flash "lite" (3.5), 3.8 is for the Pro users (the ones paying the full membership)...

Daily usage: normal task, modify this document, do this do that basic office task... anything that requires precise context is doomed for any google-based llm, that is a well known issue of even the gemma models. Only time it has been useful is while using antigravity that makes sure everything has context on every turn therefore wastes more tokens than Fable on the same prompt. Any other LLM will do it faster, burning less tokens and with better results.

About to say the same... even with the "Gemini AI Plus" plan that should be renamed to "Gemini Flash" because you are only allowed to use "the mangled model" using it for something as basic as taking notes and saving to google keep or even google docs, heck yeah a simple "add this to my calendar" completly fails in gemini... Gemini sucks for everything even its own google apps... Claude and Codex even DeepSeek and Kimi are milestones better than anything Google has launched.

It is unfortunate that having such an amazing engineers that are capable to solve any leetcode exercise in 5 minutes on the spot with a white board, they aren't able to even deliver something useful on the AI space.

Kinda FUD article... the reality is that common problems are going to be easy because the solution is probably inside the training dataset, the challenge should be adapted to make LLM's useless for example once at Defcon CTF the problems were for an unknown CPU architecture based on octal that required to write even your own disassembler... this are the kind of things that will probably be hard for frontier LLM's

There are no contextual bias, the goal of the prompt is very explicit and not about probabilistic patterns, but about the models transformer layers dynamically assigning greater weight to words like "meters" (distance) than to other tokens in the prompt.

This should be fixed in the reasoning layer (the inner thoughts or chain-of-thought) were the model should focus on the goal "I Want to Wash My Car" not the distance and assign the correct weight to the tokens.

This is another pointless article about LLM's... vibe coding is the present not the future, the only sad part of all of it is that LLM's is killing something important: code documentation.

Every single documentation out there for new libs is AI generated and that is feed again into LLMs with MCP/Skills servers, the age of the RTFM gang is over sigh

I didn't understand what is the point of the first challenge? anyone care to explain me what is the objective/task because there are just two circles and basically no instructions, kinda feels like a real job (lol)

I'm a full stack developer and still don't know what supabase does other than being a firebase replacement...

Like seriously are they seriously trying to sell us how difficult is to setup a simple nginx server? Not sarcastic just an old developer that hasn't get this new fad of going "serverless".

Yes but their primary target audience is... Teens and young adults and maybe some p.dos looking for "ASMR sutff" if you now what i mean.

What that target audience have in common... tons on times and low to zero income, so since they have time to spend trying to not pay because they don't have money to begin with, then it will happen.

Twitch is betting on having a "YouTube moment", when those teens and young adults become adults with real money, they would love to avoid the hassle of going ad-free for 10 bucks a month because they love the content that is there, so for them would be at max 1 hour of their total income, meanwhile it has to stuck taking the loses till that moment happen.

I have a BR700G it works flawlessly with apcupsd, minimal to no configuration... the cheap units of APC is for really really desperate consumers, the battery didn't last more than a year from my experience like the BR* ones that usually last 3 years, these lines is the better of the both worlds cheap and sophisticated enough to work "smart" but as expensive as the SMT/SMC units.

Also like everyone say... remember to look for pure sine wave output if you are going to "protect" servers, for network equipment you could use the pwm based ones.

Only if you have carpal tunnel problems or painful arthritis, it will make you happy and as productive as a normal developer with vs/vscode/jetbrains/etc

The amount of time you need setup *vim in order to make it a useful editor is not justified, i have seen people waste weeks on it and the only reason is if you have a condition that vim helps to masquerade.

Personally m waiting to get enough money to justify a kinesis keyboard, that will the only reason to improve my neovim knowledge... as for the moment the vim keybindings for vscode are as good as neovim.

You are not getting the point... it is not about the ratings or win ratio, is about how fast you can climb the ladder to 1200 elo (like OP suggested) from 100 at chesscom/liches, 9 months is unrealistic without paying the "pay-to-win" fee, 2 years is more realistic timeframe to get that elo.

If you are think m wrong go start a free account and prove me wrong, you have less than 9 months to go up to 1500 elo.

Thats what my chess teacher used to put me in their local chess club tournaments, in order to get a FIDE Elo, i would have need to go (on that time 1995~1999) to official tournaments and that was expensive for a rural area...

M not saying m on the level of masters but definitely not a newbie... either way i don't need proof just go to chesscom, open a new free account and start grinding... you are gonna see fast the pattern on how the platform gets you engaged but you need to look careful, it entices you with free wins before going hard on you, having three to five games a day was the eureka moment to get good ratings... either go slow or pay premium to get an edge on the game and preferred match making.

Learned chess while young... elo around 1200~1500, pickup the game at 2023 because everyone has been talking about chess, maybe related to the netflix series but since i don't have netflix account and couldn't care to watch about a young-love drama series with chess around it, it wasn't the main motive.

Anyway opened a chesscom account, i got completly destroyed by online people, elo 100 according to chesscom, move to bots could beat 1200 up to 1500 elo eaisly... decided to give a try to online matches again destroyed but then started to find the pattern...

chesscom free account give me "three fair matches" that i could win to engage me in the platform, then it pairs me with either bots, 3-second chess engine cheaters or chess streamers playing with smurfs accounts, these pattern repeat almost every time sometimes is three, other times is four even five, if i play three games on 1 day the next day i would have another "three fair matches" and so on... so whats my point?

Unless you are paying good money for the subcription you are pretty much "meat canon" for others, the same applies to lichess... if you play 3 games a day for at least 2 years months you could get the 1200 elo, if you pay-to-win or pay the premium you would likely reach that 1000 elo more eaisly because is convenient for them to no match you against known cheaters on the platform...

IMHO... Django hold a high standard in terms for projects running beyond the infamous 5+ years of support... still have some projects that with minimal changes (mostly configs and dependencies that got redundant/integrated into django -like choices-) are working with the latest version, surely 5.0 will not be that drastic if you are already doing software "the django way".

Things that are Django achilles heel are not developing software "the django way", mostly anything that needs some client-side to work is a PITA, requires lots of rewriting and custom templates that at some point you gotta start looking django as a backend more like the full stack framework that was meant to be. Also anything related onto getting things to production is another PITA that hasn't been solved or touched in years, dx deployment is one of those things almost any JS Framework out there is doing things better than what django supports.

are we talking about copilot or chatgpt? chatgpt4 provides more value than copilot since i can use it for other things, copilot is just for "code" and doesn't provide any value other than autocomplete prediction to me...

Maybe if you are website designer... ohh m sorry i should say the modern way website designers call themselves today "frontend engineer", maybe if you are a designer it could help you with all the boilerplate your react/astro/svelte requires just to add a <table> with two columns.

Maybe if you live around the bay area where you get paid absurd money by the hour for a job that can be done overseas for 10 bucks, outside or the rest of the world that 100 usd is per week tops, thats way too much i would pay for service that only barely helps my in my day to day...

If you need to constantly use chatgpt/copilot to do your job properly, maybe its time to learn something else because your job will be automated sooner or later, what will be more cheaper a 100/hr or the 20/month service.

I would gladly pay for copilot 5 USD a month (its an specific use case IA), but the current price point 20 month is incredible expensive it only add costs if the developer you are hiring requires it to properly function... thats the only value it adds.

IMHO If you are a code monkey copilot makes you a 5x developer, 10x if you have 5+ years of experience on the business

Maybe selling tokens/search could be benefitial for copilot, like only pay for what i use will be an interesting approach.

And thats why it isn't worth, you are adding complexity to a problem that doesn't need to be complex, jsDoc is the best of both worlds, type safety, zero complexity and it works with typescript (if you still really want to use it).

Personally i think typescript isn't going to be around much time on the frontend side, maybe on the backend by some people that still believe in adding another compiler layer to their build time.

When you want to integrate a javascript module that isn't annotated with types... types in a dynamic language like javascript are complex once you start doing really abstract things like generics the fun of typescript ends, also there is a recent push to abandon typescript mostly fueled by svelte fan base but nevertheless typescript is really complex, easy to shotgun yourself.

Depends a lot on what you are doing but mostly stylelint is the one that make me look on configuring eslint/stylelint for prettier...

https://prettier.io/docs/en/integrating-with-linters.html

> 1. Use npm. Better use pnpm, faster for those that don't have 1 Gbps uplink and live outside the usa.

> 3. Use Typescript Better use jsDoc, d.ts files are scary and are really complex, you need a typescript wizard la matt pollock just to tell you are mistakenly trying to concatenate a string and int.

> 5. Use prettier yes use prettier but don't forget to configure eslint to work with prettier otherwise you are going to have a bad time.

> 6. Use react Use whatever fits your problem but if you are going to really use rect then do it with Next.js, no point in still using CRA/CRACO

> 7. state management Avoid using state, try to solve every problem with pure functions, if you must use state then consider redux/react-query

> 8 & 9. You should still recommend stylelint, specially if you are going to write css by hand otherwise don't just use tailwind, tailwind is bootstrap with extra batteries (change my mind).

Main reason i hate freeCodeCamp videos is that the vast majority of them are made by Indian tutors and there is something weird with the indian pronunciation of the english that just feels like someone making a parody video more like sounding professional.

You have pretty much described the "php way" of developing things, is funny js-driking cool-aid guys talk about "SSR" being the thing to have, when we already had that, we already found that it can't scale without lots of money and started pushing computation to the client and now we are bring them back because it's not fast on the initial load...

damned we have come to full circle whats next, an assembly language so we can run binary code inside the browser kinda the way activex and applets just used do it... wait isn't that the whole point webassembly? sight you f** javascript developers have done it again sight

> But I just hope to learn to stop feeling bad for honest mistakes.

A: By trying to fix them.... this is the way.

It is simple, the difference with people that take feedback personally and the ones that don't, is that the last people are willing to make an attempt to fix the problems the others don't want to be bother.

I always take personal feedback with this mentality "ok fine, lets do it your way", if it works and its either better or equal than my solution then fine good for you, if it isn't then its the awkward/refreshing? moment when you have to tell them they were wrong and you have proof.

ChatGPT et al has run online coding exercises/trivia/exams pointless imho, leetcode problems are cool solution to the problem if you want someone with lots of "raw abstract knowledge" but zero real world experience.

Promotion in USA seems to be more like a thing that most happen every one or two years at work rather than for work well done, last time i had a friend that was working for a faang (pre-layoffs) and show me he was like a third level manager for several projects because he had 3+ years at company, he still does the same job as when he got in just that now has to manage or mostly deal with upper management request for estimated time to finish every day.

So in the event that i lost everything, i mean catastrophic, like my house burned to the ground with all my belongings, i have no kin nor "trust alternate people" configured for my account, my password manager requires my "synced in google/apple drive/cloud" passkey or my last known device, i can't retrieve it in anyway, how can i recover my account?

Either have to prove that m me to my account provider, which essentially is huge security hole since what data it will be required to prove might be more easy to fake (kinda like how people do sim swapping) and stole my passkey or do the "crypto thing", that if you lost your decryption key all your money is gone forever and ever and start fresh.

I mean my point is... password are not going to be deprecated, we had so many attempts to murder them but their convenience outmatch any other solutions, feels like passkey aren't well designed imho if the backup requires a password, then passwords won't be deprecated... maybe passkeys aren't meant to replace password but long-sessions oauth tokens if you ask me why passkeys exists.

This looks and feels like passwords with extra steps...

I mean now i need to "store, manage and secure" my per-user-certificate sorry "my passkey" myself and if its get compromised its my fault, how are passkeys more "secure" than enforcing a secure long password that the user can't change unless he met certain conditions and its conveniently stored inside the password manager i just built.

What happens if i lost all my devices due to a fire? at least a password let me still access things we these i might need to prove again that i am me, it might even be more easy to steal accounts because i can ask google to change it because i lost my passkey.

This is interesting around my area, there was a blackout at 7am to 7.30 am UTC-6 so could these be related?

Quite difficult to review that many hours of video but from the syllabus there are probably some missing pieces, mainly stored procedures and triggers, yes i know MySQL sucks for that particular job and probably you didn't touch them because planetscale probably doesn't support them (not a planetscale dev but this is my wild guess).

Also didn't see any topic on choosing the right charset/collation for the right data and why.

I would rephrase the question... what InfoSec jobs doesn't involve spending writing documentation?

pentesting? 20% finding the low hanging fruit, 80% writing and explaining your findings.

forensics? 10% finding how they did it, 90% writing and explaining your findings.

malware/policy/security/cloud security analyst? 100% writing and explaining your findings.

the list goes on and on... you are basically and a slave for word processing software, thats why totally understand OP quitting infosec.