/tech/ - Technology

>>29206
>"let's get this severely unwell man to scream the n-word because it's funny."
This really ought to be criminal. And to make such a thing as he did while ill is very inspirational and impressive. Wish he could have gotten help without losing anything too important to him, or having to persist in the way he did.

>>29206
I enjoyed watching him but once people started trolling him nefariously and getting kicked out of his house things started to go down hill fast.

Saint Terry also gave us the glowie meme

Yep. You take away the schizophrenia and we would be talking about Terry in the same breath as Bill Gates and Steve Jobs.
Which funnily enough also means we would almost certainly abhorr him like we do the other two.

>>29267
He became legend either way, he had a legendary soul not even mental illness could stop him.

>>29267
Steve Jobs had no computer skills at all.

i cant take him seriously because im constantly getting #based edits of him saying the uygha word while the caption calls him a genius, and that its proof geniuses are racist or something.
tech wise is temple really that impressive, or if anybody spent as much time as he did on it could they come up with something similar?

>>29284
It's impressive because no-one would rationally exert the effort required to create what he did, it's a work of Art more than anything

>>29284
Technically putting together a rudimentary OS is something that every systems programmer kind of knows how to do
He just became chronically unemployed due to schizophrenia and ended up being able to devote a ton of energy to making TempleOS
The display and audio are pretty rudimentary. He claimed these limitations were because of precise instructions from "God", which was probably just a psychotic filter over nostalgia and lo-fi stuff being easier to work with

>>29267
>You take away the schizophrenia and we would be talking about Terry in the same breath as Bill Gates and Steve Jobs.
No we wouldn't. Those are bourgeois entrepeneurs. Terry actually knew what the fuck he was doing.

>>29284
It's not something that asks for a lot of skill but you have to be at least decently competent as in have a good technical education and some work experience. The actually impressive thing is that he did it all alone, which takes a ton of time. The reason you don't see people doing what he did is that it's kind of a pointless endeavour that needs a lot of free time and skill.

Sorry for the doublepost.

>>29206
it isn't impressive at all, why do technically illiterate people keep repeating this. any second year student can write something similar or superior in less time. case in point go to https://wiki.osdev.org/Projects most things there are several steps ahead of templeos and were made in a much shorter time span

terry had 2 things going against him: he had a severe mental illness that drastically reduced his efficacy as a programmer, and he was old. he was from a time when programmers weren't expected to be as competent (or specialized?) as they are today. zoomers, when they dedicate themselves to low-level stuff like osdev or tooling for other programmers, are unironically the best programmers because they have been educated under much more rigorous standards: they understand their code doesn't only have to work, it has to be efficient, readable and maintainable

open source projects from 25 years ago were practically cryptic compared to the nice repositories with well thought dependency trees, standardized build systems (cargo, cmake, meson, etc.) and predictable development practices (version control, issues, commits, pull requests, tags, documentation, etc.)

>>29284
imagine if a guy raised sheep, sheared them, washed and processed the wool into yarn, spun the yarn into threads and then made his own shirt. but the shirt smelled bad because he didn't wash it properly and it was uncomfortable because the cut was wrong (no tailoring skills)

if you have only played videogames and jerked off your whole life, you will probably think that despite the shortcomings this man was very impressive. on the other hand, if you have ever been to the countryside you probably know old ladies that could have done a better job, in less time and maybe even with worse tools

>>29292
I'm not deep into the specifics of templeOS, but afaik the impressive thing about terry's projects was his custom compiler toolchain which eventually became the HolyC jit compiler. Most single-person projects on osdev are bootloaders/exokernels or just another unix clone and who can blame them for not wanting to leave the comfort of gcc? Bootstrapping a novel x86_64 system with its own C-like language, multitasking and otherwise slightly-above-dos capabilities is actually more impressive next to this, though i agree people seem to overestimate his abilities by a cringe-inducing degree.

>>29292
> any second year student can write something similar
No, not any. Writing an OS from scratch, expecially something with a bare minimum of functionality is hard. Not unique, but hard amd time consuming. And I say this while having been friends in highschool with a dude who wrote his OS (with like 4 rewrites)

>>29296
>>29297
I don't know what to tell you, the cmu curricula includes a compilers course on the second year and from the recorded classes I have seen it goes way beyond the baby first c compiler that is the holyc jit

>>29292
> they understand their code doesn't only have to work, it has to be efficient, readable and maintainable
Maybe not that. Once upon a time hardwere was simply much more limited, and as such it demanded to write more optimized code to get the thing to run at an usable pace at all.
> open source projects from 25 years ago were practically cryptic compared to the nice repositories with well thought dependency trees, standardized build systems (cargo, cmake, meson, etc.) and predictable development practices (version control, issues, commits, pull requests, tags, documentation, etc.)
Absolutely. Look no futher than the Bash source code: it's an utter disaster. C and friends aren't nearly strict enough to force people to actually compartmentalize and document their shit, let's not even talk about the eons of advancement we had in the build tooling.

>>29298
A JIT compiler is indeed an impressive feat for a single person, more than a language or OS as of now

>>29298
>baby first c compiler
I don't know if the jit compilation is notably advanced or simply some kind of threaded execution, but it's certainly not the kind of pcc-clone you would write after reading the dragon book.

>>29206
>>29296
Oh, and i also remember he used this hyperlinked multimedia file format called DolDoc. A nice touch i'd say.

>>29300
>>29301
it depends. if you build the os around it (no virtual memory or protection rings) then it isn't that much different from a regular compiler: you simply output the binary to the next address to be pointed by the program counter. the reason you usually only see jit compilers for bytecode platforms (java, lua, etc.) is because it would be hard to implement this on a modern os that does implement those features

it is the same with incremental compilation: it would be easy to implement in templeos/holyc, but it is hard to implement on modern systems, so it is usually done on top of a virtual machine

If there are so many programers on leftypol why do we never see any of their projects? It's hard for me to imagine there being programmers with enough skill here to critique Terry Davis for any reason other than to cope. Further doubt most second years can program an operating system and compiler when most operating systems courses concern merely how to interact with a UNIX operation systems.

>>29304
>you simply output the binary to the next address to be pointed by the program counter.
TempleOS doesn't have memory protection, yet it supports multitasking and thus needs some degree of linking/indirect adressing associated with that, right? And it's not like a POSIX program doing the same on W|X pages would represent a notable leap in complexity.

>>29307
>needs some degree of linking/indirect adressing associated with that, right?
no. you just have to check that your programs don't overlap. they can and they will if you make a mistake

>represent a notable leap in complexity
it does because it makes it hard to change the .text section while the program is running (in principle you shouldn't be able to)

>>29308
>you just have to check that your programs don't overlap
I had a look at the code of a game (from https://archive.org/download/TempleOS_ISO_Archive/TempleOS_V5.03/Tos_distro_2017.11.20_19-52.iso) and it doesn't appear to be loaded at a fixed offset, the same program can be run as an arbitrary number of tasks. No virtual memory only means each program has to reside in core (TOS ostensibly implements timesharing and threading without multitasking).
>it does because it makes it hard to change the .text section
You would more-or-less have to load object code into W|X marked memory, which comes with the territory. Write protection on the .text segment only makes writing self-modifying code harder.

>>29309
>and it doesn't appear to be loaded at a fixed offset
you don't need fixed offsets to overlap. all programs and all cores share the same address space. the kernel and drivers do have fixed offsets, programs are allocated with malloc (often right next to each other), but because they share the address space you can overlap. here is terry explaining it himself

>No virtual memory only means each program has to reside in core
in main memory, it also means no protections, which would be considered a giant flaw in any consumer os but it is a feature in templeos. the point is that it isn't as impressive as you think, any compiler could do something similar in a similar environment

TempleOS BTFO's the linux kernel, Windows NT does, any project on osdev does

>>29306
I do systems more the programming nowdays