Disappointed to see continued increased pricing for 3 Flash (up from $0.30/$2.50 to $0.50/$3.00 for 1M input/output tokens).
I'm more excited to see 3 Flash Lite. Gemini 2.5 Flash Lite needs a lot more steering than regular 2.5 Flash, but it is a very capable model and combined with the 50% batch mode discount it is CHEAP ($0.05/$0.20).
No South park hum? But Rick and Morty is there so it's not a policy against foul-mouthed cartoons; Smiling friends is also missing but perhaps it's too unknown or too recent to be there.
If you sort the top 50 best rated episodes of all time on IMDb you would already have 90% or the quotes people search for, so I would focus on your those if you ever try to take a stab at it.
I've been able to find four-leaf clovers easily since I was a kid. Like the OP they seem to just "pop out" at me as I quickly scan the ground, but I do not have the same background with synesthesia.
Since four and five leaf clovers tend to grow in patches, I transplanted a patch I found and have kept it in a planter on my deck so that anyone who has never found a four leaf clover can find one!
I'm interested in how these old IDEs were used during the transition from assembly to high level languages. It seems especially topical given the LLM integration into today's IDEs.
Back then was it common to have a split or interleaved view of high level and assembly at the same time?
I'm aware that you could do something like the following, but did IDEs help visualize in a unified UI?:
$ cc -S program.c
$ cat program.s # look at the assembly
$ vi program.c # edit the C code
A quick search shows that Borland Turbo C (1987) had in-line assembly:
myfunc ()
{
int i;
int x;
if (i > 0)
asm mov x,4
else
i = 7;
}
From the 1987 Borland Turbo C User's Guide [0] "This construct is a valid C if statement. Note that no semicolon was needed after the mov x, 4 instruction. asm statements are the only statements in C which depend upon the occurrence of a newline. OK, so this is not in keeping with the rest of the C language, but this is the convention adopted by several UNIX-based compilers."
The Turbo Pascal version of this was even better, because you could do a whole block of asm instead of just a single line at a time. As I remember there were annoying limitations in the C version around labels and such. It was incredibly useful when writing performance oriented code at that time because it was very easy to write code that would outperform the compiler.
I'm pretty sure you could do asm blocks in Turbo C using brackets, e.g. asm { ... }, though it might have been a Turbo C++ thing (i never really used plain TC much).
It was not very common to interleave assembly in MS-DOS IDEs. Assembler and its IDE were separate tools you paid for. But not unheard of.
You could "dump" your OBJ file for assembly.
Later C compilers got some better inline assembler support but this was towards the 32-bit era already.
Also Borland had its own compiler, linker and such as separate binaries you could run with a Makefile but you really never had to, as why would you when you can do that in the IDE in a single keypress.
It was quite common actually, most DOS C compilers supported asm{} blocks and Turbo Pascal also supported inline assembly. Paid assemblers like MASM were high-end tools.
On Unix though it was more common to have .s files separately.
I was writing game engines in Turbo C and assembler in 1988. I don't remember using inline assembler until the 90s. I just had all the graphics routines in a separate .asm file which was part of the build process and then linked in.
Unifi G4 Doorbell Pro [0] is a great self-hosted option. I've been very happy with mine over the last year, but I was already bought into the unifi ecosystem with a UDM Pro SE and U6 mesh APs.
Alternatively, the Reolink Doorbell cameras for anyone who doesn't want to be on the Unifi/Ubiquiti platform. Also I believe all these cameras provide a generic RTSP feed which can be consumed by any computer running Frigate[0], an open-source NVR/AI platform.
If it matters to you, Reolink is a Chinese owned company. Not passing judgement one way or another, but if avoiding Unifi over the remote incident matters, I could see this factoring in as well.
Separate vlans for iot devices with strict firewall rules is generally enough to mitigate the threat of most iot devices phoning home i think. We’re already in the territory of hobbyists who should be able to manage that with these suggestions like ubiquity and frigate.
I have a Unifi Protect system and am generally pretty happy with it. The biggest problem I have is similar in spirit to the Ring problem: remote access is required to get push alerts from the mobile app.
I have the same UI protect system w 7 cameras. After that incident, I made it local only. I like Unifi gear, but I became super paranoid after that incident.
After some digging around, I found homebridge [0] (with the homebridge-unifi-protect plugin by hjdhjd [1]) which fixed that for me by tying the UI Protect system into Apple's HomeKit ecosystem (which also leverages the homekit secure video that keeps alerts/motion/snapshots on iCloud). Now all our devices are able to have it popup alerts for motions, packages, etc.
It's not perfect, but this way I'm able to get alerts without tying in to Unifi's SSO system. I also still like to open the UI Protect app when I'm not on the local network to sometimes archive videos, view cameras, mess with one of the new UI PTZ cameras, so I have backup access options, including Tailscale. Tailscale doesn't give me the alerts I want, but lets me access the app as if I were still at home. I also have it tied in with HomeAssistant and recently began playing around with go2rtc.
I'm a super-newb when it comes to all this but 2022 is when I began getting fed up with all these privacy nightmares and began to teach myself selfhosting, docker, etc so I can mitigate all this. Unfortunately, I'm the only one who knows how to tinker and keep all this updated. However, I do have documentation for my wife how to access everything and start fresh to make it easier on her by using UI's SSO way so it "just works" as they say in the Macintosh World, when I'm no longer around.
I haven't seen any in depth performance comparisons, but I had also started to dig into this and I was leaning towards using Scrypted instead of the HomeBridge plugin. Just wanted to name-drop them for others also interested in all this
Scrypted is faster when I tested it a year or so ago.
I almost jumped into their Scypted NVR system (with all the bells and whistles all fully integrated in one app and would cost me $70/yr for 7 cameras) but it seemed like I'm trading one remote access system for another hence why I decided to keep using Homebridge (HB) until the HB project starts becoming stagnant or unless something breaks on my end. You can have Scrypted+HomeKit just like I have Homebridge+HomeKit setup, but Homebridge is pretty fast for me on Apple Silicon hardware so as they say "if it ain't broke, dont fix it."
This is the beauty of open source projects like these. Pick which works for you and just keep it updated and donate a few $ when you can to support the project.
One more vote here for Scrypted. I have a bunch of cameras from various vendors, some with open FW, some with their original FW, all cut off from the internet. They used to be connected to Frigate but due to performance issues I offloaded the work to Scrypted on a RPi and an AppleTV and the setup works great. Bonus, it’s easier to set up and use compared to Frigate, especially for the less technical people in the family.
I've had the Vision Pro since launch, and the only thing that keeps me coming back to it nearly daily is the Mac Virtual Display for my MacBook Pro.
It's just so useful to have a huge display wherever you want it - no hunching over looking down a small laptop screen. This is especially useful on a plane where I'm not even able to open the laptop completely due to the tight space.
My main gripe is: Why do I need a separate Mac at all?
Even the original M2 Vision Pro has more than enough horsepower to run the virtual Mac inside of the headset, so it seems like a fake limitation.
I'm looking forward to it being lighter weight and smaller, and for them to make the Mac Virtual Display native to the Vision Pro experience without the need for a separate computer.
The only use case I’m interested in is long-haul flights. How comfortable is it to wear for an extended period? Do people interrupt you to ask you about it? Are you able to use all the features sitting in economy while gesturing?
Expectations vary but I find that regardless of perfect comfort and battery life I don’t want to go more than a few hours without a break. I’ll watch a jumbo movie or a few hours of tv on a flight no problem
Comfort: For me, the answer was “no” out of the box. I had to exchange my light seal twice before I got the right one. This was no charge but obviously it would be great to get it right the first time and it seems like a large number of users never did this and those are the ones who complain most about comfort. Not blaming the users btw, Apple needs to fix this if they care about the product
Onlookers: I always wonder if somebody will bug me and they never have. If you are self-conscious, it might not not be the product for you
Gestures: You learn that they can be done very subtly and you don’t need to bother the person next to you
Overall yes (with some caveats below), it's comfortable enough to wear for a few hours. I'm usually in the window seat and I try to be discreet, so nobody has really asked me about it.
All of the features work well even with no internet (for linking the mac & vision pro). Travel mode takes care of issues with the plane's movement and the sensors work excellent for gesture / tap detection even in complete darkness, and there's never a need for big sweeping arm gestures with the vision pro - you can operate it in a small space.
I bought the belkin headstrap to help distribute the weight, and I imagine the success of that type of accessory is what prompted apple to introduce the new dual knit band. With that extra support I can wear it for a few hours but I definitely have goggle marks on my face for a while afterwards - like you would have after skiing.
Most people don’t care, the only weird/funny part is that they assume you can’t see with it on. I’ve never been interrupted and I wore mine on a flight just a few months after it was released.
Apple won't do that because it means creating a Mac enclave[0] inside of what is supposed to be a secure OS. Apple wants the Mac to be firewalled off from the rest of their product line because the Mac has root access and other things that let the owner tamper with device security. To be clear, you can keep such a device secure, but Apple believes the additional work to support keeping such an owner-controlled device secure is "working for free".
To elaborate on that last bit: macOS ships with a number of utilities and frameworks specifically to detect and remove known-malicious software. macOS also has to operate notarization infrastructure for supporting non-MAS apps, as well as boot infrastructure to deliberately run untrusted or known-insecure OS kernels. None of their other platforms need this[1], because they have strict code signing enforcement. The web browser and developer mode aside, the only code that is ever allowed to run on device is code written by an entity with a business relationship to Apple. Anyone who wants to ship malware has to create a paper trail and expose themselves to getting a legal ass-ramming.
Of course, in practice the enhanced security of iOS and its derivatives is really just an excuse to extract 30%, a price most app developers aren't willing to pay. Putting all your "real apps" behind a virtual display that you have to carry a separate device around for is a way to contrive an inconvenience whose answer is "ship a native visionOS app."
I suspect the whole reason why Mac mirroring is even a thing at all is because Apple realized iPad apps weren't going to cut it on a $3500 VR headset, and this was their quick hack to make the Vision Pro useful while they figured out a way to browbeat their developers into officially supporting it. A task which, by the way, has failed miserably.
[0] No relation to "exclaves" - i.e. bits of security-sensitive code that have been isolated from regular iOS system processes and run inside SPTM's "secure kernel" domain, but can still IPC back and forth. The most likely approach to "Mac enclaves" on visionOS would be enabling Hypervisor and shoving macOS in EL1.
[1] EU DMA notwithstanding - in fact, much of Apple's anger regarding the DMA boils down to the fact that complying with it while keeping their devices secure means shipping macOS-like antimalware infrastructure on other platforms.
author here - I made this comment elsewhere, but I still think apple made the right call even if it leads to a bit of confusion at first.
As others have pointed out, sliders have limits & knobs don't, so I do think they have their place on touchscreens.
If a digital knob needs to be turned several times (e.g. 1080º, common in DAWs), the "default" way to interact with a knob on a touchscreen - circling again and again - is slow and uncomfortable. Adding "slider" gestures on top of the default behavior is a nice way to perform many turns quickly and easily.
I'm curious - what UI mechanism would you use instead?
Those gestures have been pretty standard with mouse movements in music software for a while. Apple brought it to touchscreen, but they didn't invent the multi gesture knob interface.
I had a Novation Zero SL mkii in 2008 which had a "universal knob". It acted as an HID mouse input and when you spun the knob it would click your mouse and drag it vertically. It worked with most music software at the time because vertically dragging a knob had become fairly standard.
I'm more excited to see 3 Flash Lite. Gemini 2.5 Flash Lite needs a lot more steering than regular 2.5 Flash, but it is a very capable model and combined with the 50% batch mode discount it is CHEAP ($0.05/$0.20).