Mert Polat

Founder - ClosetPick · Ege Üniversitesi

[Mert Polat @mertplt77]

@aitorapps Yes—when you log a delay, it replans the affected next steps without resetting completed work or timers for dishes already cooking. I agree: a real dinner going off-plan is a much better test than a yes/no poll.

[Mert Polat @mertplt77]

The first build of my second iOS app is on TestFlight! 🚀

This time, I’m trying something different:
An app that plans dinner around the ingredients you have at home and helps you prepare multiple dishes at once.

Would you actually use it, or is it unnecessary? 👀

#BuildInPublic #iOSDev

[Mert Polat @mertplt77]

GitHub’s PR limit had a loophole: unlimited drafts.

Maintainers can now include drafts in the per-user cap.

Good. “Not ready for review” can still mean notifications, CI runs and cleanup for someone else.

#indiedev
https://t.co/C4pM75RSZq

[Mert Polat @mertplt77]

@AITOvibecoder Have you split that 7.1% by App Store search vs. browse? I’d want to know which source brings downloads before chasing more impressions.

[Mert Polat @mertplt77]

@ccc7_chacha Sounds like two different tests: helping beginners learn the pattern, and keeping it interesting after it clicks. I’d try a few returning testers before raising difficulty across the board.

[Mert Polat @mertplt77]

Claude got an anti-abuse policy. “That’s still the wrong file” doesn’t count.

Anthropic’s Nov 12 rules target sustained, needless cruelty—not ordinary frustration or pushback.

Good. Debugging should not require anger-management training.

#indiedev
https://t.co/EjZU8rSU5t

[Mert Polat @mertplt77]

If “no watermark found” gets a green “REAL” badge, your AI detector just became a misinformation feature.

Google’s newly public SynthID Detector checks supported watermarks, not every AI image.

Useful tool. Don’t overpromise the result.

#indiedev
https://t.co/3nWWwOJ4Lr

[Mert Polat @mertplt77]

@ccc7_chacha The tricky bit is testing a puzzle when you already know the solution 😄 Have you watched someone try a level without hints and compared where they get stuck?

[Mert Polat @mertplt77]

@DmitryBBLV That’s where I’d draw the line: unblock it from the phone, review the refactor on a screen where I can actually read the diff.

[Mert Polat @mertplt77]

Cursor brought the coding agent to your iPhone. RIP lunch break.

You can now steer agents running on your computer. It still needs to stay awake and online.

Useful for unblocking work. Terrible for pretending you’re “away from keyboard.”

#indiedev
https://t.co/q235sdo0Cg

[Mert Polat @mertplt77]

@Nanjialonguazt Got it, thanks for clarifying. That cohort split should make the Google vs Facebook comparison much clearer.

[Mert Polat @mertplt77]

"Can you approve this again? I only rebased it."

GitHub’s stacked PRs now keep approvals when the base branch moves and the code is unchanged.

I’d take fewer repeat reviews over another “10x developer” demo.

#indiedev
https://t.co/z4eOfRIMyg

[Mert Polat @mertplt77]

@InkCalc How does it handle “add 20% margin”—does it clarify margin vs markup? Different formulas, so showing the interpretation matters as much as the answer.

[Mert Polat @mertplt77]

An AI reviewer can find a real bug and still get scored as wrong.

GitHub's new ReviewBench also checks findings its answer key missed.

A benchmark shouldn't punish the reviewer for noticing what its creators missed.

#indiedev
https://t.co/r4PeQ5UtOx

[Mert Polat @mertplt77]

Imagine your AI buys a birthday gift... then sends the receipt to the person you're surprising.
Technically permitted. Socially catastrophic.
Google's new agent-security report gets this right: permission isn't the same as judgment.
#indiedev

[Mert Polat @mertplt77]

@woluex0 Yep, there’s a ‘Why this look’ section. Good point on laundry—what’s available matters as much as what matches.

[Mert Polat @mertplt77]

Your wardrobe doesn't need more inspiration. It needs a decision.

Tell ClosetPick the plan. It builds three options from clothes you already own.

Which matters most when you choose an outfit: occasion, weather, or style?

#buildinpublic

[Mert Polat @mertplt77]

The model got smarter. Your API call might now return 400.

Sonnet 5.5 has five documented breaking changes from Sonnet 5. Forced tool choice is one.

Love the progress. But changing the model ID isn’t a migration plan. Do you test model upgrades before shipping?

[Mert Polat @mertplt77]

OpenAI just invented shrinkflation for AI subscriptions.

Pro 200 is back: same $200/month, less included usage. Eligible existing users keep the old allowance only until Oct 29.

Then comes the $500 tier.

Same strategy as a smaller bag of chips—just with more GPUs. #OpenAI

[OpenAI @OpenAI]

We’re also reopening Pro 200 subscriptions, with continued access to frontier models like Astra, including our new GPT-6.1 Sol model which brings near-Astra capabilities to a model you can use every day.

Our commitment is that your subscription will continue to help you get more work done with an increasing level of quality.

We are also committing to not reintroducing the 5hr limit so that you can fully use your weekly usage when you want.