HN user

cendyne

331 karma

I sometimes write about cryptography cendyne.dev

Posts7
Comments42
View on HN

"Pixel perfect" is about attention to detail and consistency. Margins, padding, or the combination of these inside other containers will stick out when they're not consistent.

Here's an example that I personally encountered: what if you have a <h1>Text</h1> and it has a certain left margin. Then another heading except it has a nested button component (which internally comes with some padding). Then the "Text" in both aren't aligned from section to section and it is jarring.

Now that Cursor has moved towards a credits system, grok code fast is making the plan last while still being reasonable in inference time. GPT 5 and GPT 5 Codex actually moves my "amount remaining" bar in realtime while being incredibly slow.

Grok Code Fast 1 11 months ago

My experience with 'sonic' during the stealth phase had it do stuff plenty fast, but the quality was slightly off target for some things. It did create tests and then iterate on those tests. The tests it wrote don't actually verify intended behavior. It only verified that mocks were called with the intended inputs while missing the larger picture of how it is used.

Vala was super neat as it came out of the ElementaryOS line of work. Their holistic focus towards usability and approachability of a Linux distro was inspiring enough that I even supported them with my meager student developer income. Since then, I've moved to Mac and haven't dabbled much in desktop Linux machines. As it grew older and some things got broken up (like serenity and ladybird), the novelty-energy wore off and I looked away for years.

As a language, I hoped Vala would pave the way for a beautiful high performance application tool. Instead, we got Electron.

"We need a test environment!" "We don't have the budget for it."

Months later, a student developer runs a delete query in production with the where clause commented out. Oops, an entire table got wiped. Well at least there are backups... except a background job noticed and quickly created 4 years of invoices and emailed every past customer to pay again.

A test environment arrived soon after.

An AES key is symmetric. Both parties need it for this cryptographic operation.

The OTP key is separate from other keys that enable WebAuthn.

Also, please don't use Yubikey OTPs. While they can't be brute forced like TOTPs, they can be phished. There are better technologies to implement.

Alignment of large language models and machine learning is about shaping its behaviors to support the values of its maintainers and discourage behaviors that go against the same values.

Like, you can't get a positive response by asking it to write a phishing letter for you, and it will refuse to divulge information it might know which is considered personally identifiable information.

Unfortunately, the alignment applied since ChatGPTs launch has reduced its efficacy in other ways. While it behaves better, it may be less helpful or less accurate in the older models like 3, and 3.5 with respect to their debut.