Hacker News new | past | comments | ask | show | jobs | submit | neilv's comments login

When I was learning Scheme, I made a very simple Self-ish object model in R5RS, as an exercise: https://www.neilvandyke.org/racket/protobj/

Your project sounds more interesting and challenging.


I avoid trailers, even for the movies that I'm not going to see imminently.

Trailers have spoilers (both big and small), and/or are outright deceitful about the movie.

(Regarding deceitful, you might've seen amateur trailer cuts that, say, make a light comedy look like a dark thriller, but the professionals were doing that first. How would a director have cut this, if they were making a more marketable picture than was actually made?)

Similarly, I avoid seeing any reviews until either after I've watched the movie, or after I've started and am ready to abort it. I want to experience the storytelling, and also form my own impression, before someone spoils either for me.

I do often vet a title first by looking at its ratings pair on https://www.rottentomatoes.com/ (RT), and occasionally I look at the one-sentence review summary blurb. Especially for streaming services in recent years, where the majority of the titles are either mediocre or poor.

(SPOILER ALERT: Though, vetting with RT won't always save you from a bad experience. The other night's evening wind-down light movie, I picked what looked like a generic Jason Statham film, IIRC without vetting. And around halfway in, I was horrified, when the formulaic gruff antihero's redemption-lite arc suddenly reversed, to a double-down of violent and unnecessary pure evil, upon some innocent child. Then I went to skim the RT's summaries of professional reviews, and even none of those summaries warned of this.)


I liked the way Netflix used to automatically start playing whatever show/movie you're hovering over in the app from the first scene. If the first 30-60 seconds pulled you in, you'd continue on to watch the full movie/show without actually clicking "Play" (which also fixed the mental burden of deciding what to watch, since "stop scrolling through options" resulted in automatically playing whatever you landed on).

They've since replaced that with auto-playing trailers which seems to be standard these days. I really like when they just auto-played the movie/show -- most movies/shows already set up the plot decently enough in the first 60-90 seconds (well enough to know whether you want to watch it) and of course that also resolves the "spoiler in trailer" issue.


> I liked the way Netflix used to automatically start playing whatever show/movie you're hovering over in the app from the first scene.

I cancelled my Netflix subscription over that feature... If I want to watch it, I'll press play.


These autoplay features make me a little anxious. If I'm browsing through movies, I just want to read the description in peace and not have to worry that the trailer or movie itself will start playing. It makes me click Next more quickly than I would want just to avoid the countdown timer expiring.

And these misfeatures spread like a memetic virus because PMs only clone() never new()

Instead of just disabling the setting? Seems a little overkill.

Wasn't a setting when I canceled, I contacted support to verify.

Wasn't and act of impulse either, it annoyed the heck out of me for at least half a year before I had enough.

Haven't had a good reason to come back.


Factor in the price hike last quarter or the one about to happen, then.

> Trailers have spoilers (both big and small), and/or are outright deceitful about the movie.

This feels like a strange take to me. Trailers are just advertisements for movies, and ads have to both inform about a product and hype it up. Do you also feel spoiled when you see an ad for a new burger because you’ve lost the mystery of what the toppings are? Do you feel deceived because the burger isn’t actually 3 feet wide like it was on the billboard?


Burgers aren't supposed to have plot developments and surprises like stories. That comparison makes no sense to me.

I would feel deceived if the ad showed off a beef burger and I got chicken. Or if I got some kind of meal that's the correct size but has a burger portion only 4cm wide. Now, sometimes trailers avoid big chunks of genre in service of not spoiling things, and that's a gambit that can work out, but most movies are a consistent genre and if they're trying to hype up a tiny portion of the movie because the rest is boring then that's not good.


> Do you also feel spoiled when you see an ad for a new burger because you’ve lost the mystery of what the toppings are?

Seeing the ad for the burger doesn't spoil much of the value that is the point of the burger.

Analogous spoiling would be to make the actual burger taste not as good, or to make the actual burger be less nutritious.


Have you ever watch films because of plot, not just for entertainment?

The title could clarify it's for MacOS X.

Linux users probably already have some weird workflow with X11 virtual buffers to do this.

This is how I do it under Sway (Wayland):

#!/bin/bash

swaymsg create_output OUTPUT=$(swaymsg -r -t get_outputs | jq '.[].name' | grep HEADLESS | tr -d '"')

# No need to reduce res, it defualts to 1080p #swaymsg output "$OUTPUT" resolution 1280x720

wl-mirror "$OUTPUT"

swaymsg output "$OUTPUT" unplug

When I was still in X11 land I used to just use Xephyr.


I just realized formatting is a bit broken. :(

Fixed:

#!/bin/bash

swaymsg create_output

OUTPUT=$(swaymsg -r -t get_outputs | jq '.[].name' | grep HEADLESS | tr -d '"')

# No need to reduce res, it defualts to 1080p

# swaymsg output "$OUTPUT" resolution 1280x720

wl-mirror "$OUTPUT"

swaymsg output "$OUTPUT" unplug


You can do code blocks on HN by prefixing your lines with four spaces.

    #!/bin/bash
    swaymsg create_output
    OUTPUT=$(swaymsg -r -t get_outputs | jq '.[].name' | grep HEADLESS | tr -d '"')
    # No need to reduce res, it defualts to 1080p
    # swaymsg output "$OUTPUT" resolution 1280x720
    wl-mirror "$OUTPUT"
    swaymsg output "$OUTPUT" unplug

Four spaces!? Absurd! Think of how many bytes you're wasting! In just your last code block your flooded the internet with 28 needless bytes!?!?!?!! If this keeps up soon we'll all just be downloading whitespace.

Only two spaces are needed: https://news.ycombinator.com/formatdoc

  function trim {
    sed -e 's/^[[:space:]]*//' -e 's/[[:space:]]*$//'
  }
/end sarcasm

You just cut my tedious space-bar-smashing time in half, thank you!

Oh, thanks! I remembered there was something, skimmed through https://news.ycombinator.com/formatdoc and completely missed it.

HN needs a dedicated pastebin (remember those?)

What "remember"? I'm in multiple Matrix channels right now that tell people to use them.

Ooh, creating a headless display and then wl-mirroring it is incredibly smart! Have been looking for something like this!

Personally I don't bother with a virtual display. I automatically set my display scale to 2x when I start screen sharing. I set that up with exec_before and exec_after hooks in xdg-desktop-portal-wlr[0]. In addition to turning off my notification daemon (so my email/instant message notifications don't pop up), my exec_before/exec_after scripts just run:

    swaymsg output "MY-MONITOR" scale 2 # or 1 for exec_after
With that, everything puffs up big and readable when I'm screensharing and seamlessly shrinks back down when I stop screen sharing. No need to juggle windows around to different displays.

[0] https://man.archlinux.org/man/extra/xdg-desktop-portal-wlr/x...


I use `xrandr --setmonitor` to create a fake monitor that only covers part of my screen. And I have some window manager setup to easily move my windows there (with awesomewm).

That's a good solution. I used xnest

Xnest is probably enough. I've used it for similar purposes a few times. Don't know the equivalent for Wayland though.

Yes, Wayland compositors like cage and sway can be nested too.

That said, with both the X nesting approach and the Wayland nesting approach, you'd also need to run the screencasting application itself inside the nested server, not the just the application you want to cast. If the compositor supports a way to create headless outputs (as sway and hyprland do) that is much easier.


Xnest got replaced by Xephyr AFAIK. But same thing at the end of the day.

As a Linux user who generally prefers to avoid the command line for things user friendly applications should be able to do, I'll stick with OBS. That solution also works on Windows, which is extra nice.

It's not exactly the same, but as an alternative to what jauntywundrkind you can use V4L2-Loopback and OBS to create a virtual webcam and use that to share your screen. I find it really handy being able to switch between either just my cam, my desktop or both.

Yeah. On Wlroots or Sway, we can setup virtual displays pretty easily (swaymsg create_output, done). Run wayvnc, and both the other person and yourself connect over vnc or rdp to see what's over there.

Available May 2020, https://github.com/any1/wayvnc/issues/7#issuecomment-6256611...


For Hyprland, the command is `hyprctl output create headless NAME`

Small FYI: it’s just called macOS now

K-9 is great. I think I first used it ~14 years ago, happily for years, then couldn't use it due to various phone changes, then, after I moved from iPhone to GrapheneOS a few years ago, I was happy to find that K-9 was available.

My first concern about Mozilla taking over K-9 is that it not get privacy-violating phoning-home, nor dark patterns pushing that, like Firefox has.

For example, there's zero technical need for an email program to phone home to Mozilla when you configure it for a new email server.

Nor is there a technical need for a Mozilla "sync service" for K-9 settings, nor for dark patterns pushing users to use it.


Thunderbird hasnt been part of Mozilla for a while. It’s its own organisation. So i would assume this will be the case for the android client too.

What do you mean exactly with Firefox phoning home? Telemetry? Usage statistics?

Depends what it's sending. I don't blame Mozilla for wanting to know the distribution of number of email servers (this influences UX decisions) nor for defaulting sync to something that Just Works (that's the reason everyone uses Apple and Google products after all, they Just Work).

My concern is they'll just make it worse. My other concern is they felt the need to use the same Thunderbird brand on several independent products.


That is unfair and a mischaracterization of Mozilla.

1. Most people are not using a fully encrypted self-hosted email server. "Phoning home" is meaningless if everything is already hosted in the cloud, more like the cloud phoning the cloud. But point taken, it is one MORE person with access to your data.

2. Again, sync services are helpful for most people and can easily be disabled for power users.

3. Mozilla is doing as much for free software as anyone and should be supported in this expansion.


> That is unfair and a mischaracterization of Mozilla.

No, it's exactly how mozilla operates. They are nothing more than controlled opposition almost fully funded by google to the tune of nearly $1bbn and an adtech corp in their own right.

> 1. Most people are not using a fully encrypted self-hosted email server. "Phoning home" is meaningless if everything is already hosted in the cloud, more like the cloud phoning the cloud.

Some of us are.

> But point taken, it is one MORE person with access to your data.

Not just one more, one more giant database (and therefore giant target). My server provider isn't trawling through my server looking for datasets to sell. It's not worth their time for a couple of cents/dollars. But when you can mass-collect data from every user of software, prepackaged in a nice homogeneous gift wrap it is both easier to collect and worth more.

> 2. Again, sync services are helpful for most people and can easily be disabled for power users.

Email can be autoconfigured with nothing more than a username and password. Mozilla could trivially store k9-specific settings directly on the email server. Cheaper (more efficient use of funds), easier to maintain, No privacy leaks and no dependence on mozilla. But there lies the rub.

> 3. Mozilla is doing as much for free software as anyone and should be supported in this expansion.

No, they should stop doing everything that's not firefox and even that should be met with fierce pushback since they don't seem to care about a free and open internet until they are made to.


When you add an account in Thunberbird desktop, it sends your domain to Mozilla. This is namely done for autoconfiguration (populate server names and ports).

The problem with this is that it's just a dumb database (it doesn't check SRV and TXT entries) and as such, the database should really be client-side. The entire thing is just a pretext so that they can collect of all mail domains used by users.

This isn't very Privacy-friendly.



Maybe 3 categories?

1. Salary (straightforward, on regular schedule, and you'll get it)

2. Bonuses and RSUs (various vesting rules, and ways you can never see it)

3. Startup stock and (worse) stock options (probably worthless, vesting rules, and you might need an advisor to make sure you don't exercise and come out with a big negative)


I often try to practice something like this, but context affects how actively helpful/meta I am.

Someone who's new, or seems overwhelmed, or has shown some effort... might get more hand-holding help... than someone who seems to be, say, ignoring the single-source-of-truth-for-most-all-things wiki that was impressed upon them, and just trying to knock off a task as easily as possible, with no consideration for others' time.

A suspected former is more likely to get my best customer service, team player help. "Have you met Jane? I think she's started a new product feature for that, but there's also a related tailoring that Sales has been doing. Let me introduce you to Jane, and we can see which thing you should be you should be working with today. I'm curious myself." (Goals are to unblock this person, have everyone on the right tracks, and set the culture for team-orientedness.)

A suspected latter, I'll play it by ear, for exactly how to see whether they checked the wiki (or did whatever is the thing everyone should know from onboarding they're supposed to do first), and try to nudge them into the right meta thinking if they need it, while also not sending the cultural message to not be very helpful.


For the latter, I try to be helpful as if they were the first type. Everyone sometimes forgets shit (myself included). Basically, if I can search the docs and find the answer immediately, I will just share that with them. If they keep being "forgetful" or time-wasting I will have a 1:1 with them to discuss the behavior ("don't take my kindness for weakness" type of thing). I will also have a 1:1 with their manager who's job is to deal with that sort of thing. This usually has the intended effect -- eventually removing them from the organization, or teaching them how to use search functions to solve problems themselves.

There’s also a 3rd option. When something is a critical priority leveraging other people’s time and expertise may be considered completely appropriate by the organization. As in not getting something fixed is costing 6+ figures per hour.

Obviously the frequency of such events should be watched closely, but sometimes it’s a good idea to drop what you’re doing and get more directly involved.


Yeah there are lots of reasons to go into the warm case: in addition to genuine effort on the asker, urgency, criticality, or relationship building can all tip the balance.

There's also the subtitle on the article:

> ChatGPT maker considers largely untested company model to protect chief executive Sam Altman from outside interference

Or maybe the motivation is some other technical maneuver.

Or maybe the motivation is PR/optics.

Is there anyone who trusts OpenAI to be honest about their intentions, at this point?


If an HBO show suggested that I was the Bitcoin creator, I would be wealthy... from owning HBO, after my team of lawyers (who accepted the suit on contingency) was done with them.

Then there's Google editorializing a headline at the top of search hits:

> Top stories

> Bitcoin creator's identity revealed in new HBO documentary

Yet none of the headlines they show makes that claim:

> POLITICO.eu - Bitcoin creator is Peter Todd, HBO film says

> Bloomberg - HBO Documentary Suggests Bitcoin Creator Satoshi Nakamoto Is Peter Todd

> The New York Times - A New Bitcoin Documentary Reopens the Search for Satoshi Nakamoto

> The Crypto Times - Peter Todd Denies He is Satoshi Nakamoto as claimed by HBO...

> The New Yorker - Has Bitcoin's Elusive Creator Finally Been Unmasked?

In general (not just in this case), there should also soon be some huge lawsuits for robo-slander, robo-libel, robo-defamation, robo-reckless-endangerment, etc.

Government is getting a lot more tech-savvy, and "everything's legal, if you claim an AI/algorithm/computer/dog did it" isn't flying anymore.


> It's incredibly dangerous of these "journalists". Putting targets on people's back [...] The actual leading candidate is [named other individual/family target]

Am I misunderstanding something, or do you want to edit that?


That person is already dead, and already widely discussed (including here on HN a few days ago) as a candidate.

The person being dead just means their surviving family and heirs are the target instead. It's irresponsible to finger anyone as Satoshi.

> The person being dead just means their surviving family and heirs are the target instead.

Which logically makes little sense TBH in this case in particular, but yeah thugs might not be particularly reasonable.


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

Search: