HN user

fireattack

1,898 karma

fireattack#outlook.com

Posts4
Comments671
View on HN
news.ycombinator.com 1y ago

Tell HN: Google Search shows irrelevant results for me for days

fireattack
2pts0
chuttenblog.wordpress.com 5y ago

Why are there so many Canadians in India?

fireattack
141pts122
news.ycombinator.com 6y ago

Ask HN: Why using “system font stack” when you can just use “sans-serif”?

fireattack
3pts1
news.ycombinator.com 7y ago

Is Twitter's “like” function currently broken?

fireattack
2pts0
I watched Gemini CLI hallucinate and delete my files 12 months ago

Windows doesn't either.

    >move 1 ..\1\
    The system cannot find the path specified.
            0 file(s) moved.
But the issue is you can't ensure LLM will generate the command with trailing slash. So there is no difference in Windows or Linux for this particular case.
0ThreadHN
Lost Chapter of Automate the Boring Stuff: Audio, Video, and Webcams in Python 1 year ago

playsound

This library is unfortunately effectively abandoned -- it hasn’t received any updates in over four years, and its latest version doesn’t work at all: https://github.com/TaylorSMarks/playsound/issues/101

(A workaround exists: downgrading to version 1.2.2, but that comes with its own issues.)

The last time I experimented with audio in Python, I was surprised by how lacking its multimedia libraries are.

For example, when I needed to read audio files as data, I tried `SoundFile`, `librosa` (a wrapper around `SoundFile` or `audioread`), and `pydub`, and none of them was particularly satisfying or has seen much active development lately.

If you need to read various formats, pydub is probably your best bet (it does this by invoking ffmpeg under the hood). I was hoping for a more "native" solution, but oh well. Unfortunately, `pydub` is also unmaintained and has some serious performance issues (for example: https://github.com/jiaaro/pydub/issues/518 )

0ThreadHN
A wonderful coincidence or an expected connection: why π² ≈ g 2 years ago

Totally unrelated to the content, but about the website itself.

The site completely breaks when I visit it. After some investigation, I found out that if I enable Stylus (a CSS injection extension) with any rules (even my global ones), the site becomes unusable. Since it's built using the React framework, it doesn't just glitch; it completely breaks.

After submitting a ticket and getting a quick response from the Stylus dev, it turns out that this website (and any site built with caseme.io) will throw an error and break if it detects any node injected into `<html>`.

[1] https://github.com/openstyles/stylus/issues/1803

0ThreadHN
How to use GitHub Copilot in your IDE: Tips, tricks, and best practices 2 years ago

I'm Copilot customer since day 1 and I'm mostly satisfied with it, until a few months ago.

I noticed a mini project I was working on suddenly stopped showing any autocomplete results. The log shows the API returns empty `model deployment ID: []` constantly.

Upon investigation, I found the reason is that my code includes the word "ass", because it is a project about parsing .ass subtitle format (Advanced SubStation Alpha).

I have to rename all my variables to work around it, and eventually this isn't enough: I have a subprocess.run to call ffmpeg to burn subtitles on video when encoding. And the VF (video filter) for that is called.. ass.

Here is an MRE (try typing anything after the line `subprocess.run` and observe the copilot log):

    import subprocess
    from pathlib import Path
    STORAGE_DIR = R"G:\test"
    def process(f, subtitle, output):
        subprocess.run([
            "ffmpeg", "-v", "error", "-stats", "-i", output, "-vf", f"fps=60,ass={subtitle}", "-c:a", "copy",
            '-c:v', 'h264_nvenc', "-pix_fmt", "yuv420p", output, "-y"
        ])
    def main():
        for f in Path(STORAGE_DIR).iterdir():
            process(f, f.with_suffix('.txt'), f.with_name(f.stem + '_ouptut' + '.mkv'))
    if __name__ == '__main__':
        main()
Other people have reported similar issues [1] about other "bad" words. I mean, I get it, but don't they think having "ass" in that "profanity blacklist" would cause way too many false positives? To make it worse, MS/GitHub seem refusing to make any solution or even communicate about this issue.

I've opened a ticket on GitHub's support system too, but got no response whatsoever after weeks. I'm very disappointed as a paid user.

To any GitHub employee if you're by chance reading this, it would be tremendously helpful if you can help mention this issue internally. Thank you in advance.

[1] https://github.com/orgs/community/discussions/55630

0ThreadHN
gh-116167: Allow disabling the GIL 2 years ago

Would GIT affect how Python runs across multiple Python processes?

I'm asking because I encountered a weird phenomenon before.

I use a simple Python lib called "schedule" which is to run some tasks periodically (not precise). And I often run a script multiple times (with different arguments) to monitor something say, every 30 seconds. So they're in three separate Python Interpreter processes.

What I've noticed is that while when I initiated them, they were something like 5 seconds apart, they eventually will end up running in sync. Probably not related to GIL at all, but I guess do no harm to ask.

0ThreadHN
Cracking Meta's Messenger Certificate Pinning on macOS 2 years ago

Yes, if the user want to disable all the protections and choose to install malware it's their choice. You can already do so on *nix, Windows, and macOS (albeit more complicated). Not sure why a phone OS would be different.

Your line of thinking is basically "think of the children".

0ThreadHN
Bugs I've filed on browsers 2 years ago

Oh God, after reading your post, it suddenly struck me that maybe I should try using a different Google account, even though I don't really have anything special in my main account.

I registered a new account, and voilà, it suddenly worked!

I compared all the options and couldn't find anything that would cause this until I discovered that my main account's Tools > Accessibility Settings had "Turn on screen reader support" enabled, for which I have no idea why.

I turned it back off, and the issue immediately disappeared.

To be fair, I think enabling this accessibility option shouldn't cause such a bug; but for now, I can finally use it again.

Again, thank you for taking the time to test it!

0ThreadHN
Bugs I've filed on browsers 2 years ago

I can reproduce it with Chrome, Firefox, both with no extension.

On three computers.

In case it was not clear, here is the full STR.

1. Click on a cell. 2. Click on the formula bar. 3. Type something, Press enter (you can even directly press Enter).

Expected: the next cell is highlighted but not in edit mode. So you can use arrow keys to move highlight around.

Observed: the next cell is in edit mode; you cannot move arrow around.

To make it worse, sometimes (but not always), if you press enter AGAIN, it would go to the next (the third one) and STILL in edit mode; you need to press enter AGAIN, which makes the highlight move to the fourth cell, only then it properly exists edit mode.

The bug is also reproducible with Esc (cancel change):

1. Click on a cell; 2. Click on the formula bar; 3. Directly press Esc.

Expected: exit edit mode

Observed: you are still in edit mode. You have to press Esc again.

A gif demo: https://i.imgur.com/KV6rRP6.gif

0ThreadHN
Bugs I've filed on browsers 2 years ago

I like to file bugs about any software or service I use, too. Just did a quick check, under my current email I have 101 bugs filed for Chromium and 91 for Firefox (which I no longer use extensively, so they're mostly years old). Since I'm more of a power user than dev (I don't work in software or CS field) my "hit rate" isn't as high as the author, but I'm still proud about it.

One unfortunate thing is that browsers are probably one of the only remaining major services/applications (that are not directly developer-oriented) which you can report bugs and actually get some tractions, or even get them fixed.

Try to report a bug about MS Word. Or Spotify. Or Google Maps. I've tried all of them and it's safe to say nothing happens no matter how obvious the bug is or how easy it would be fixed. I don't even know if the "feedbacks" I sent ever reached the developers.

For example, recently Google Sheets has a bug that if you input data in the formula bar (instead of directly in the cell) and then press Enter, it somehow enters the edit mode of the next cell, instead of just highlights it like if you edit in the cell. This is driving me crazy and I've filed multiple feedbacks about it, but I have no hope to see it getting fixed any time soon.

0ThreadHN
uBlock Origin works best on Firefox 4 years ago

In general, I don't find any noticeable performance issues using ubo on Chrome, but again I'm the "install it and forget it" type.

I do encountered a weird specific performance bug that is when cosmetic filtering is enabled, some heavy sites become very lagging, even affecting typing [1][2]. It also happens with other ad blockers. Unfortunately the author was unable to reproduce it.

[1] https://github.com/uBlockOrigin/uBlock-issues/issues/1687

[2] https://bugs.chromium.org/p/chromium/issues/detail?id=129245...

0ThreadHN
Why you can’t buy a thermapen on Amazon (2016) 5 years ago

I buy from Amazon purely for 30-day return policy which is quick and painless to process (single click, drop it to UPS store, done. Don't even need printer, tape or box if the original package it still intact.)

Can't say I have the same experience from other ecommerce websites, brand ones or aggregated ones like BB or Newegg.

0ThreadHN
Simple bank shutdown goes awry, leaving customers without account access 5 years ago

Just in case I wasn't clear, I think we both agree that message being unauthentic is a different matter from the incident itself. Just that angry people often can't/wont distinguish the two.

I pointed this out because GP seems to not be affected, so his view won't be obscured by this; but the phenomenon should be noted.

0ThreadHN
Simple bank shutdown goes awry, leaving customers without account access 5 years ago

I don't know. I do like their statement since it's direct and not vague as you said.

However, I've seen multiple times on Internet that a company made an apology which looked fair to me as a bystander, while people who were affected screamed "sorry but not sorry".

I think it makes a big difference emotionally depending on if you're involved or not.

0ThreadHN