Hacker Newsnew | past | comments | ask | show | jobs | submit | kristopolous's commentslogin

Finally I can get those 100 free hours I was promised

so you know how to swap audio with -map without having to look it up?

I do, yes. Though that's not really the point, it'd already be enough to know where to look it up.

no the point is that there are some things I've done a hundred times and I never remember it because it's designed in a wildly bad way. ffmpeg, gpg, openssl and git has those things all over the place. Is it -c:v or -v:c? I don't know. used to be -vcodec so it's -v:c now? no it's -c:v I think because they swapped it?

There isn't internal consistency to really hold on to ... it's just a bunch of seemingly independent options.

The biggest problem is open source teams really don't get people on board that focus on customer and product the way commercial software does. This is what we get as a result


> Is it -c:v or -v:c?

Sure, I agree with all of this. Like I said above, the syntax (and, even more, the defaults) isn't great. I'm just arguing that "improving the syntax" should not mean "hiding complexity that should not be hidden", as the linked project does. An alternative ffmpeg frontend (i.e. a new CLI frontend using the libav* libraries like ffmpeg is, not a wrapper for the ffmpeg CLI program) with better syntax and defaults but otherwise similar capabilities would be a very interesting project.

(The answer to your question is that both -vcodec and -c:v are valid, but I imagine that's not the point.)

> The biggest problem is open source teams really don't get people on board that focus on customer and product the way commercial software does.

I believe in this case it may be more of a case of backwards compatibility, with options being added incrementally over time to add what was needed at the moment. Though that's just my guess.


ffmpeg doesn't go away. it's still there. people can use tig and git, having something that isn't insane can live in harmony with the other thing.

I was actually writing, been doing it full time for months. I've spent probably over 1,000 hours ...

Not trying to make any money, just feel compelled to do this.

A fiction story about how personal computers have dismantled society over 40 years... it takes place in 1983 and involves a vulnerable opportunistic time traveler who's getting more than he bargained for.

Here's some quotes to give you a feel:

"The smartphone is the electrical stunner in the slaughterhouse of society"

"You’ll be able to access any TV or radio station in real time, around the world, talk to people overseas in high resolution video with live translations for free and be bored by it"

"In the future the hermetic spaces of solitude will be breached as we build a global village. The private will become public and, ironically, the public will become private as the streets empty of experiences taken indoors, inside of bedrooms, beneath our screens of glowing grace."

It's intentionally meant to be ambitious, brutal and challenging. And hopefully insight will materialize from the dust of forgotten dreams.

If you are interested in reading it, just hit me up


Interested. Sending an email

this is probably the best tool for this stuff now: https://github.com/deepbeepmeep/Wan2GP

It has fastwan ... probably will have this soon. it's a request in multiple tickets: https://github.com/deepbeepmeep/Wan2GP/issues


also don't ignore webp and avif ... those can really do wonders.

In any media, people only see what they want. There's a psychological term for this, Motivated Reasoning.

https://en.wikipedia.org/wiki/Motivated_reasoning

If you want to break this you have to know the person and ask key questions afterwards. Their distortion field is held together by beliefs and principles, not empirical analysis.

For instance, for my father, the question "how is this treating people responsibly? How can we expect the behavior of those guards to be held accountable?" would pierce this ... but really you have to know how the person doing motivated reasoning thinks.


His Dad will be smart enough to know these questions are trying to set him up. Maybe try having a real conversation and not trying to change his mind. After all, there is a good chance you will be that Dad in the future (no matter how hard you tell yourself you won't be). Tell me how I now.

I'm almost 50. I won't be. I have friends who are becoming grandparents now, still no interest.

I have half a century of talking with my father. If you think this is my first strategy as opposed to one that took years of therapy and personal struggle, I dunno what to tell you.

There's a wide body of social and psychological research on this stuff including multiple university departments (communication, psychology, sociology, management, teaching, etc) because "simply talking to people" doesn't actually work.


Thanks kristopolous. We have a very similar story (I'm a few years older). I think I'm at the "I've given up point" because his glee at others' suffering is just too painful to even address. So: he get's hellos at holidays and that's it.

People have discovered being an open sewer spewing hate and prejudice gets likes, views, reposts and advertisers

It's also a very easy job. You don't need to do journalism, be diligent about citations and accuracy, use robust analysis or careful language.

You don't even need a script. Just hop on a hot mic, blame an oppressed scapegoat and see money roll in.

The content is evergreen, trivial to create and performs great!

Just like you don't have to be a doctor to swindle people with phony medicine or a psychology degree to hustle people as a psychic.

The problem is we've taking the smooth talking performative palliatives of these slick mountebanks and christened their confidence games as sacred free speech instead of the hatemonger hustle it is.

And unfortunately, like Albania’s Nationwide Ponzi scam of the 1990s, these crimes have become institutionalized power and their bullshit is bringing the country down with them.

Other than personal gain, what ought be the consequences of arsonists shouting "fire" on the crowded Internet?


Real conversations cannot involve one or more persons trying to change another's mind?

It's a very prevalent form of cynicism, which I find ironic because in high school every student learned to write persuasive essays, but "adults" like to tell each other not to change people's minds. It's a subtle meta-rhetorical move used to undermine rationalism and formal education.

So does this apply to every single person all the time?

Nothing does.

It's about successful communication of authorial intent.

60 Minutes is not trying to say "Justice Served!" and shake pom-poms here. But, someone could read it that way, and it would be unintended.


I use local models + openrouter free ones.

My monthly spend on ai models is < $1

I'm not cheap, just ahead of the curve. With the collapse in inference cost, everything will be this eventually

I'll basically do

    $ man tool | <how do I do this with the tool>
or even

    $ cat source | <find the flags and give me some documentation on how to use this>
Things I used to do intensively I now do lazily.

I've even made a IEITYuan/Yuan-embedding-2.0-en database of my manpages with chroma and then I can just ask my local documentation how I do something conceptually, get the man pages, inject them into local qwen context window using my mansnip llm preprocessor, forward the prompt and then get usable real results.

In practice it's this:

    $ what-man "some obscure question about nfs" 
    ...chug chug chug (about 5 seconds)...

    <answer with citations back to the doc pages>
Essentially I'm not asking the models to think, just do NLP and process text. They can do that really reliably.

It helps combat a frequent tendency for documentation authors to bury the most common and useful flags deep in the documentation and lead with those that were most challenging or interesting to program instead.

I understand the inclination it's just not all that helpful for me


This is a completely different thing to AI coding models.

If you aren't using coding models you aren't ahead of the curve.

There are free coding models. I use them heavily. They are ok but only partial substitutes for frontier models.


I'm extremely familiar with them.

Some people, with some tasks, get great results

But me, with my tasks, I need to maintain provenance and accountability over the code. I can't just have AI fly by the seat of its pants.

I can get into lots of detail on this. If you have seen tools and setups I have done you'd realize why it doesn't work for me.

I've spent money, the results for me, with my tasks, have not been the right decision.


> I'll basically do

    $ man tool | <how do I do this with the tool>
or even $ cat source | <find the flags and give me some documentation on how to use this>

Could you please elaborate on this? Do I get this right that you can set up your your command line so that you can pipe something to a command that sends this something together with a question to an LLM? Or did you just mean that metaphorically? Sorry if this is a stupid question.


Yes, I use simonw's `llm` for that: https://github.com/simonw/llm

Example:

    $ man tar | llm "how do I extract test.txt from a tar.gz"

I'm not the OP, but I did build a tool that I use in the same way: https://github.com/scottyeager/Pal

Actually for many cases the LLM already knows enough. For more obscure cases, piping in a --help output is also sometimes enough.


i guess op means: $ man tool | ai <how do I do this with the tool>

where ai could be a simple shell script combining the argument with stdin


Is your RAG manpages thing on github somewhere? I was thinking about doing something like that (it's high on my to-do list but I haven't actually done anything with llms yet.)

I'll get it up soon, probably should. This little snippet will help you though:

   $ man --html="$(which markitdown)" <man page>
That goes man -> html -> markdown which is not only token efficient but also llms are pretty good at creating hierarchies from markdown

I bet you could do the same thing with pandoc and skip serializing to HTML entirely.


Not the OP, but I did release my source :D https://github.com/scottyeager/Pal

My tool can read stdin, send it to an LLM, and do a couple nice things with the reply. Not exactly RAG, but most man pages fit into the context window so it's okay.


I use llm from command line too, time to time, is just easier to do

llm 'output a .gitignore file for typical python project that I can pipe into the actual file ' > .gitignore


> My monthly spend on ai models is < $1

> I'm not cheap

You're cheap. It's okay. We're all developers here. It's a safe space.


While I say this somewhat in jest, frugal is just cheap but with better value.

> I'm not cheap, just ahead of the curve.

I'm not convinced.

I'm convinced you don't value your time. As Simon said, throw $20-$100/mo and get the best state of the art models with "near 0" setup and move on.


this is the extent to what I use any LLM - they're really good at looking up just about anything, in natural language, and most of the time even the first hit, without reprompting, is a pretty decent answer. I used to have to sort thru things to get there, so there's definitely an upside to LLMs in this manner.

Have you looked at tldr/tealdeer[0]? It may do much of what you're looking for, albeit without LLM assistance.

0: https://tealdeer-rs.github.io/tealdeer/


No not at all. Let's do it this way:

What percentage of people who drive drunk have consumed thc within the time window of blood detection?

Now this is a more reasonable number

I really don't know why this number is significant. Accidents are situational and people who engage in situations where accidents are more frequent likely make other decisions about consumption and lifestyle which involve things like cannabis

Who cares?

You aren't going to elevate the behavior of the population by regulating a plant



Well it doesn't disclose when it was first updated.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: