Shubham Yadav

Keep Going |
Journey to #100DaysOfCode |
SE |
Full Stack Developer

[Shubham Yadav @CoderOO7]

@Manvi_World Answering questions like these 🤔

[Shubham Yadav @CoderOO7]

@SahilExec It should be D) Both rate limit and memory issues

1) A simple workaround is to hit the req in batches (100 at a time with a delay or wait till completion)
2) Use vertical scaling via process clustering, to let your machine multiple cores handle incoming reqs.

[Shubham Yadav @CoderOO7]

@naturedotcom None as such, but I used NewPipe most of the time for 🎵

[Shubham Yadav @CoderOO7]

@SahilExec 1) Revert Immediately
It's a critical payment flow bug, so can't take any chances

2) Fix branch permissions, no direct pushes to main, PR + review required

3) Mistakes happen that's fine, will ask junior dev to fix it and do proper testing before raising pr to the main branch.

[Shubham Yadav @CoderOO7]

@suryanox7 Yes.

The AI boom only took off mostly in the last 1y. Before that, we worked without these tools with no issues.

Anyone who learned the manual way can switch back and forth easily.

[Shubham Yadav @CoderOO7]

@SahilExec - Prevent Failover: If primary goes down, a replica takes over to prevent downtime
-Traffic Distribution: Replicas handle read queries, freeing up the primary

However, accidental data corruption in primary db will reflect in replicas. Backups are still required for recovery

[Shubham Yadav @CoderOO7]

@SahilExec Go with Redis.

10k writes/sec will melt Postgres with disk I/O and lock load.

Instead, using Redis as the primary data store (zero cache misses) and Postgres as Async DB to store progress seems fine.

If I choose A over B
Need to compromise with performance, not sure on that

[Shubham Yadav @CoderOO7]

@dnfslvtt Khela to h, but yaad nhi aa rha

[Shubham Yadav @CoderOO7]

@SahilExec But if we use setTimeout with 0 delay and setImmediate together, then we can't decide priority, it depends on system which one it will pick first.

[Shubham Yadav @CoderOO7]

@siya_twt_ They are afraid of change 🤔

[Shubham Yadav @CoderOO7]

@piyush784066 ThinkPad + linux = Beast

I used to like the ThinkPad P series workstations old one

The reason is fully customisable
- Dock station
- You can easily swap the cpu, gpu. Those parts were detachable.

But now every component is soldered in the motherboard, so can't be swapped.

[Shubham Yadav @CoderOO7]

@sujithdotcode Ask yourself, as a company CEO(male) you will keep a girl as HR or boy ?

[Shubham Yadav @CoderOO7]

@Aditya_181105 Gaming

[Shubham Yadav @CoderOO7]

@manojdotdev I don't hate companies, but dumbass managers and directors in those companies a lot.

[Shubham Yadav @CoderOO7]

@SahilExec 1)F: Can slow down too
2)F: Modern NoSQL supports
3)F: Encoded (Base64), not encrypted
4)F: VM can run full OS, docker can't
5)T: Make async code looks synchronous by avoiding callback hell
6)F: Distributed locking
7)T: 2xx means req success
8)F: PK by default indexed, not FK

[Shubham Yadav @CoderOO7]

@suryanox7 That it can replace human.

With 50-75% of the web now flooded with AI content, models are training on their own output. It’s a photocopy of a photocopy. Without fresh human data, AI will literally degrade.

[Shubham Yadav @CoderOO7]

@dishasatra In this bad world there are good people too, don't lose hope

[Shubham Yadav @CoderOO7]

@ShraddhaShips I bought 1-month subscription of Claude Code for around Rs 2300. Then I later tried OpenCode and found it quite impressive. The free tier models easily handle standard coding tasks, and the other models are very affordable. For personal work, OpenCode is a solid choice

[Shubham Yadav @CoderOO7]

@debugingwithsim Then maybe it was for maintaining a private variable. Initially, there were no classes in JS. That way, by keeping a variable in the outer function and returning an inner function that does something with those variables, we can create private variables and fns too.