3 ms·

I really would like to hear perspectives of old people like your self about various recent trends and how they are a repetition of old innovations. For example

I really would like to hear perspectives of old people like your self about various recent trends and how they are a repetition of old innovations. For example example about containers (or vms?) as meta processes and kubernetes as some type of meta-os. Or computer languages which are just interfaces to the same base algorithms. Or anything really. I also have the vague feeling that there are repeating patters, but I don’t have a deep historical perspective to make this feeling more concrete. Does anyone have a blog or something?

dougabug 5y ago
Human knowledge is evolving incredibly quickly, and has since before computers were even invented. The first computer I played with was the (then new) Commodore PET, with 4KB RAM. My elementary school had one, and access to it was like some sort of special privilege. There’s a certain tendency people tend to have, which is to believe that “things were better in the old days,” or “they don’t make ‘em like they used to.” Or “there are no new ideas.” I dunno, for me going back to the pre-smartphone, pre-Google, pre-Internet world… pass. Naturally, there are a wealth of core ideas, algorithms, patterns which have stood the test of time and even evolved and flourished. But our understanding of even basic ideas has significantly changed over time. Personally, I’ve generally elected to not get too caught up with fads, and to focus on understanding fundamental ideas, algorithms, mathematics; how things fit together into robust, efficient, flexible systems.
smackeyacky 5y ago
I don't know about repeating patterns, what I see with containers is sort of novel but has its roots in competition between time sharing systems and transaction processing systems. I've never used Multics but it seems that the guys who wrote unix and VMS and other mini computer OS were time sharers. One box, lots of users, how do we stop users from stomping on one another. The final paradigm was that your programs can pretend they own the whole machine and the operating system handles keeping everybody safe from one another. This is fine if everybody runs single programs but eventually hits the wall when you need competing environments. Personal computers and networking fixed that, but now I want to share my running program, I need a server again. A big, shared machine only my environment clashes with yours, so containers are born. They are networked personal computers in conception. The other pattern is transaction programming. IBM mainframes didn't really do time sharing, you write up your job and schedule it. Some jobs are then triggered by conditions or events. Its more like a domino stack than processes trying to pretend they are the only thing running. AWS lambda functions are the modern equivalent but without the safety of proper queues that something like CICS provided. They will get there. For some problems, you need interactivity. For others, you really want transaction processing. I think transaction processing is really underutilised in modern systems as we all scramble just to make stuff work and the time sharing and personal computer modes of thinking dominate the problem space.

dougabug 5y ago

Human knowledge is evolving incredibly quickly, and has since before computers were even invented. The first computer I played with was the (then new) Commodore PET, with 4KB RAM. My elementary school had one, and access to it was like some sort of special privilege. There’s a certain tendency people tend to have, which is to believe that “things were better in the old days,” or “they don’t make ‘em like they used to.” Or “there are no new ideas.” I dunno, for me going back to the pre-smartphone, pre-Google, pre-Internet world… pass. Naturally, there are a wealth of core ideas, algorithms, patterns which have stood the test of time and even evolved and flourished. But our understanding of even basic ideas has significantly changed over time. Personally, I’ve generally elected to not get too caught up with fads, and to focus on understanding fundamental ideas, algorithms, mathematics; how things fit together into robust, efficient, flexible systems.

smackeyacky 5y ago

I don't know about repeating patterns, what I see with containers is sort of novel but has its roots in competition between time sharing systems and transaction processing systems. I've never used Multics but it seems that the guys who wrote unix and VMS and other mini computer OS were time sharers. One box, lots of users, how do we stop users from stomping on one another. The final paradigm was that your programs can pretend they own the whole machine and the operating system handles keeping everybody safe from one another. This is fine if everybody runs single programs but eventually hits the wall when you need competing environments. Personal computers and networking fixed that, but now I want to share my running program, I need a server again. A big, shared machine only my environment clashes with yours, so containers are born. They are networked personal computers in conception. The other pattern is transaction programming. IBM mainframes didn't really do time sharing, you write up your job and schedule it. Some jobs are then triggered by conditions or events. Its more like a domino stack than processes trying to pretend they are the only thing running. AWS lambda functions are the modern equivalent but without the safety of proper queues that something like CICS provided. They will get there. For some problems, you need interactivity. For others, you really want transaction processing. I think transaction processing is really underutilised in modern systems as we all scramble just to make stuff work and the time sharing and personal computer modes of thinking dominate the problem space.