3 ms·
Imagine losing all your work because a stray pointer in your word processor mistakenly pointed to a hardware register. Imagine not being able to upgrade the har
by anyfoo 6y ago
Imagine losing all your work because a stray pointer in your word processor mistakenly pointed to a hardware register. Imagine not being able to upgrade the hardware or even the software, because programs access hardware directly and directly interact with internal OS data structures. Imagine having to essentially restart the computer and go through minutes of loading time to look something up in another piece of software.
I'm a big retro computing enthusiast, and I'm not happy about the current state of abstractions (especially with web stuff), but there's definitely been advancements also.
- grishka 6y agoVirtual memory and memory protection are good things, there's no denying that. As are operating systems with clearly defined APIs, drivers, and a clean separation between kernel space and user space. But I would still say that even with those problems present, old software does feel quite a bit more reliable than the modern one. Things generally don't break themselves because they've auto-updated while you weren't looking. And UI doesn't change on its own because there's a thousand server-side toggles for A/B tests. Old software generally was made as a tool, it served its users. Modern software is made to pursue its developer's own agenda first and foremost.
- anyfoo 6y agoI'd be careful whether you're not looking at things through nostalgia-colored glasses, as humans are well known to do. Computers crashed and exhibited data loss all the time back then. PCs came with very prominent reset buttons, and when both Linux and Windows NT came up, the prospect that you could run your machine for an entire day without having to reboot was a revelation. It wasn't different on 8 bit micros: My C64 was very easy to upset, and I bet most people turned it off and back on to get it into a known state when switching programs, instead of risking that the previous software left some interfering bits somewhere. Sure, it was easy to power cycle a C64 and took mere seconds, but it was also very necessary. Well into the 90s there was a strong general attitude of having to reset every now and then just to keep things from diverging too badly. Computers crashing, hanging, and losing data was so ubiquitous that it was part of mainstream humor much more than it is now. Then, towards the end of the 90s, when things started to get better, people were so proud of uptime, that they used to brag about it in forums and chats. Nowadays, my computers, smart phones and similar devices usually run for weeks, pretty much only rebooting for software updates. That isn't always the case, but it's certainly orders of magnitudes better than the very fragile balance back then.
- grishka 6y agoDon't get me wrong, modern technology is superior in many ways. The problem I'm picking on here is that of the conflict of interest between the users and the developers. And this is absolutely not helped by the proliferation of predatory code signing and governments always siding with corporations when people try to take matter into their own hands. The problem with modern technology is that many of its upsides are compensated by software developers. You know, "disks are big, networks are fast, memory is plentiful, you don't need to count CPU cycles in $year" kind of mentality. > Nowadays, my computers, smart phones and similar devices usually run for weeks, pretty much only rebooting for software updates. On my macbook, something is leaking memory. I'm suspecting the nvidia GPU driver. I have to reboot it every week or two or else it'll start freezing even while I'm typing something.
- ddingus 6y agoInterestingly, one could run multiple instances of retro computers and their environments and get many of those advances. Just a random thought.
- grishka 6y agoActually yes, by running retro software in an emulator on a modern system you're getting mostly best of both worlds.