Post 292489 by davidpgil@floss.social
(DIR) More posts by davidpgil@floss.social
(DIR) Post #263143 by davidpgil@floss.social
2018-09-26T20:23:56Z
0 likes, 1 repeats
i feel like we should pick up where the golden age of computing left off and just go back to that thread of development, open source and free all of the hardware and software then hit resume. the current state of computing is so suffocating its maddening... this is why retro computing has been going strong lately! i am joining this this trend since this year! #retro #decentralization #foss #fosh #computing
(DIR) Post #263728 by uranther@cybre.space
2018-09-27T22:58:31Z
0 likes, 0 repeats
@davidpgil What exactly was the golden age of computing and when, in your opinion, did it start/stop?
(DIR) Post #263976 by davidpgil@floss.social
2018-09-27T23:18:02Z
0 likes, 0 repeats
@uranther I think it varies from person to person. I think it started rapidly vanishing at around the turn of century. i am still exploring when it started. im not sure why but lately i feel so bored by the cutrent state of hardware and software. all of it has become less creative and bloated. Also, things have become much more locked down by walled gardens and software dictating the hardware... it all just doesnt feel quite right to me!
(DIR) Post #264365 by davidpgil@floss.social
2018-09-27T23:21:59Z
0 likes, 0 repeats
@uranther ... i started feeling this way while trying to teach myself assembly for the 6502. it just feels like a big blue ocean of "i can be in complete control of this chip" ... theae daya things have gotten so complex that its hard for one person to just create something really cool on their own with full control over the hardware.
(DIR) Post #264366 by uranther@cybre.space
2018-09-27T23:57:22Z
0 likes, 0 repeats
@davidpgil I hear ya. It's harder than ever to gain the skills to build something useful on the computer. Google, Microsoft, and Apple would prefer it that way as they build their elite engineering teams and closed ecosystems. The rest of it is complexity that everyone is too afraid to unwind for backwards-compatibility's sake.
(DIR) Post #264398 by uranther@cybre.space
2018-09-28T00:01:11Z
0 likes, 0 repeats
@davidpgil We are pushed more and more to use computers for consumption rather than creation.
(DIR) Post #264417 by davidpgil@floss.social
2018-09-28T00:05:38Z
0 likes, 0 repeats
@uranther yep! its all getting locked down. time to reverse that.
(DIR) Post #264419 by davidpgil@floss.social
2018-09-28T00:06:05Z
0 likes, 0 repeats
@uranther yep! preach on bro!
(DIR) Post #264889 by ACE_Recliner@misskey.xyz
2018-09-28T01:09:43.462Z
0 likes, 0 repeats
@davidpgil@floss.social Is there any particular cutoff in your opinion, or is this just sentimental stuff. Personally, imo, I felt a lot of user approachability and creativity was lost in the 32->64 transition and I think that public domain 32-bit ISAs (PowerPC 1.x, SH4, etc.) will be making a huge comeback for understandable open hardware computers in the next few years.
(DIR) Post #265272 by Ricardus@mastodon.sdf.org
2018-09-28T02:01:40Z
0 likes, 0 repeats
@davidpgil Do we all get IBM System 360s?
(DIR) Post #265371 by davidpgil@floss.social
2018-09-28T02:09:54Z
0 likes, 0 repeats
@Ricardus we probably could! 😅
(DIR) Post #265377 by Ricardus@mastodon.sdf.org
2018-09-28T02:10:11Z
0 likes, 0 repeats
@davidpgil SHOULD!
(DIR) Post #265387 by davidpgil@floss.social
2018-09-28T02:11:39Z
0 likes, 0 repeats
@Ricardus but we each need to rent a studio for it 🤩
(DIR) Post #265392 by Ricardus@mastodon.sdf.org
2018-09-28T02:11:53Z
0 likes, 0 repeats
@davidpgil I'm OK with that.
(DIR) Post #265408 by davidpgil@floss.social
2018-09-28T02:09:20Z
0 likes, 0 repeats
@ACE_Recliner i agree that 32bit era was when we started losing touch. i think the limitations of pre 32bit hardware were just limited enough that it helped hackers run with their ideas while being able to figure out how the hardware worked. the abstractions were just right. yes there is some sentimentality but i do think its not JUST nostalgia. Mostly i feel like our current computing situation is pretty boring and dominated by frameworks n such.
(DIR) Post #265409 by ACE_Recliner@misskey.xyz
2018-09-28T02:13:52.918Z
0 likes, 0 repeats
@davidpgil@floss.social I would place that balance of power more in the 32 bit era (pre 3D acceleration anyway) but yeah. I think something like VRender0 built around a SH ISA could do a lot to reignite creative limited computing and allow for a truly personal canvas more then current hardware would allow.
(DIR) Post #265505 by davidpgil@floss.social
2018-09-28T02:19:45Z
0 likes, 0 repeats
@ACE_Recliner my issue with 32bit systems is that they start to abstract a bit too much. The golden age of computing for me is before 32 bit.
(DIR) Post #265506 by ACE_Recliner@misskey.xyz
2018-09-28T02:21:09.786Z
0 likes, 0 repeats
@davidpgil@floss.social Not all 32 bit systems are made equal in this regard, AmigaOS for example allowed abstraction pretty much down to the metal and so did DOS with DPMI extenders.
(DIR) Post #265565 by davidpgil@floss.social
2018-09-28T02:24:23Z
0 likes, 0 repeats
@ACE_Recliner sure. i have limited knowledge of all of this btw. im just currently obseassed with rebooting our current trajectory any way i can. i am teaching myself 6502 assembly now to get my juicea flowing.
(DIR) Post #265566 by ACE_Recliner@misskey.xyz
2018-09-28T02:26:54.121Z
0 likes, 0 repeats
@davidpgil@floss.social Cool. A logical next step from here might be 68k assembly since that'll give you more powerful machines to work with but without losing the same touch so to say (Amiga, Megadrive, Neo Geo, X68000, Atari ST etc.)
(DIR) Post #286816 by vertigo@mastodon.social
2018-09-29T15:49:12Z
0 likes, 0 repeats
@ACE_Recliner @davidpgil RISC-V is quickly becoming the RISC architecture of choice for both commercial embedded and free/libre computing designs. My own project, #Kestrel3 neo-retro computer, is built around a homegrown RISC-V implementation.PowerPC is not public domain; it's still controlled by IBM (through their stewardship at OpenPower). The RISC-V ISA, meanwhile, is under BSD license, which is one of the reasons for its relatively rapid uptake recently.
(DIR) Post #286817 by ACE_Recliner@misskey.xyz
2018-09-29T15:54:13.557Z
0 likes, 0 repeats
@davidpgil@floss.social @vertigo@mastodon.social When I say "PowerPC 1.x", I'm referring to the core 32-bit ISA pre-AltiVec which is not subject to patent or trade protections in the United States. You of course can't use the "PowerPC" trademark in connection with it but you're free to reimplement it as you wish, and projects like J2 (although for a different ISA) depend on this assumption to work.
(DIR) Post #292488 by jason@thesocialmedia.feedbackloo.pw
2018-09-27T22:39:15.746710Z
2 likes, 1 repeats
@davidpgil I love retro & obscure operating systems and architectures. I love KDE-Plasma on Arch Linux, but if I were to find the right mix of retro or obscure OS/System , I would probably switch in half a heartbeat, just to remove the stale experience that much of the internet and some aspects of modern computing now offer and have something freshOne of he more intriguing 'modern - retro' OS's that I have found is for 1980's 8bit machines with the Zilog z80 cpu thee is SymbOS (Amstrad CPC, MSX etc)http://www.cpcwiki.eu/index.php/SymbOS
(DIR) Post #292489 by davidpgil@floss.social
2018-09-27T22:44:39Z
0 likes, 0 repeats
@jason That looks seriously amazing! Love that low res look.
(DIR) Post #292490 by jason@thesocialmedia.feedbackloo.pw
2018-09-27T22:54:16.435815Z
1 likes, 1 repeats
@davidpgil I think its great, and I now find myself looking at eBay (and at carbootsales) looking out for potentially compatible machines on the cheap.The SymbOS website[0] has a links page[1] you may enjoyAlso check out the apps[2] that are available for SymbOS, its very impressive as to what 'Podatron' has achieved [0]http://symbos.de/index.htm[1]http://symbos.de/links.htm#marke1[3]http://symbos.de/apps.htm
(DIR) Post #292491 by davidpgil@floss.social
2018-09-29T23:08:54Z
0 likes, 0 repeats
@jason you blew my mind wide open with #symbOS lots of interest around me in the past couple of days... i feel like i need to join/ start this neo retro computing cult 😅 at @kragen @vertigo right? lol
(DIR) Post #292492 by kragen@nerdculture.de
2018-09-29T23:14:36Z
0 likes, 1 repeats
@davidpgil @jason @vertigo Retrocomputing is potentially fun, since it's a source of all kinds of difficult technical challenges, but what I'm most interested in here is #bootstrapping from a small base, in order to have an entire computing environment that any one person can comprehend and change in whatever way they would like, so that free software is a practical reality rather than just a legal permission. In a sense, technical challenges are the thing to avoid as much as possible.
(DIR) Post #294972 by davidpgil@floss.social
2018-09-29T16:13:07Z
0 likes, 0 repeats
@vertigo may i hear more about your retro computer? yesterday i had what i think is probably a very similar idea. i was inspired by #SymbOS and various other influences.
(DIR) Post #294973 by kragen@nerdculture.de
2018-09-29T16:21:27Z
0 likes, 0 repeats
@davidpgil Symbian?
(DIR) Post #294974 by davidpgil@floss.social
2018-09-29T16:25:15Z
0 likes, 0 repeats
@kragen #SymbOS https://en.m.wikipedia.org/wiki/SymbOS
(DIR) Post #294975 by kragen@nerdculture.de
2018-09-29T22:58:16Z
0 likes, 0 repeats
@davidpgil Oh, nice! I had no idea anyone had written a windowing system for the Z80! Can it run anything useful in 128K (like the original Mac, but with multitasking)? Or even on the 64K Enterprise 64? Can it assemble itself?I kind of wish there was an equivalent for a RISCier 8-bit microcontroller. All the embedded Z80 chips I've come across use 10s to 100s of mW, while an STM32L supposedly uses like 3.5 mW at 16 32-bit MIPS.
(DIR) Post #294976 by kragen@nerdculture.de
2018-09-29T23:00:16Z
0 likes, 0 repeats
@davidpgil The Z80 strikes me as particularly suitable for #bootstrapping because it already has a somewhat reasonable self-hosting programming environment—Turbo Pascal—which AFAIK has no equivalent on "more modern" CPUs like the AVR or ARMs like the STM32. (Do you know of anything else similar?)
(DIR) Post #294977 by davidpgil@floss.social
2018-09-29T23:04:30Z
0 likes, 0 repeats
@kragen that is something I would not know. maybe ask @vertigo ?
(DIR) Post #294978 by vertigo@mastodon.social
2018-09-29T23:39:50Z
0 likes, 0 repeats
@davidpgil @kragen RISC-V has GCC, but I generally avoid it because it's only really useful if you intend on porting and running software under Linux.I've written a minimalistic, bare-metal Forth environment vaguely modeled on PygmyForth (DX-Forth) that runs on RISC-V, but that's unlikely to be what people are looking for.Off and on, I frequently consider the possibility of writing a compiler for a language based on an enhanced version of Oberon-07. But that's just day-dreaming so far.
(DIR) Post #294979 by kragen@nerdculture.de
2018-09-29T23:41:51Z
0 likes, 0 repeats
@vertigo @davidpgil GCC can't compile standalone kernels for #RISCV?
(DIR) Post #294980 by vertigo@mastodon.social
2018-09-30T00:03:43Z
0 likes, 0 repeats
@kragen @davidpgil It can. Fedora has been ported to RISC-V, and I know there are several other distributions (usually catering to embedded applications) available as well.But, the GNU toolchain is, in my experience, completely broken if you're *not* targeting Linux.
(DIR) Post #294981 by vertigo@mastodon.social
2018-09-30T00:04:58Z
0 likes, 0 repeats
@kragen @davidpgil Put another way, if you're writing a single-address space operating system/run-time environment, GCC and GNU LD will become your worst enemy.If you're looking to port Linux or any other system that uses ELF executables, GCC is your go-to compiler.
(DIR) Post #294982 by kragen@nerdculture.de
2018-09-30T00:09:07Z
0 likes, 0 repeats
@vertigo @davidpgil Why, what happens to SASOSes?
(DIR) Post #294983 by vertigo@mastodon.social
2018-09-30T00:28:38Z
0 likes, 0 repeats
@kragen @davidpgil The toolchain (so far at least) is incapable of producing a loader format that is compatible with a single-address space.You can create binaries that are assigned specific load addresses, but that prevents you from building software where you can have multiple modules loaded at once.
(DIR) Post #294984 by vertigo@mastodon.social
2018-09-30T00:34:04Z
0 likes, 0 repeats
@kragen @davidpgil The other catastrophe with GCC is that it uses ELF, which is horrifically overcomplicated for its job. I wrote some blog articles on this topic: - http://kestrelcomputer.github.io/kestrel/2018/01/29/on-elf- http://kestrelcomputer.github.io/kestrel/2018/02/01/on-elf-2
(DIR) Post #294985 by uranther@cybre.space
2018-09-30T04:12:46Z
0 likes, 0 repeats
@vertigoI read the first article and skimmed the second so far.Are you saying that for #RISCV you pretty much don't need an executable format since it can just load the raw binary?Is there a modern format available without having to retrofit ancient specs (not that this is inherently bad)?@kragen @davidpgil
(DIR) Post #295080 by kragen@nerdculture.de
2018-09-30T04:30:05Z
0 likes, 0 repeats
@uranther @vertigo @davidpgil Any processor can just load a raw binary; that's not specific to RISC-V.
(DIR) Post #295386 by vertigo@mastodon.social
2018-09-30T05:13:29Z
0 likes, 1 repeats
@uranther @kragen @davidpgil No; the thesis of the articles is that ELF is a technological mistake on several levels.It's over-engineered, under-documented, over-specified in irrelevant matters, and is a gross violation of Unix philosophy.The 2nd part illustrates how to achieve exactly what ELF aims to support with only a small fraction of the implementation and specification complexity.ELF is complete AT&T NIH material.
(DIR) Post #297937 by calvin@cronk.stenoweb.net
2018-09-30T12:36:00Z
0 likes, 0 repeats
@vertigo @uranther @kragen @davidpgil I mean, I use a system that uses XCOFF as its binary format, that's no ELF...
(DIR) Post #298017 by jjg@social.coop
2018-09-30T12:51:13Z
0 likes, 0 repeats
@calvin @vertigo @uranther @kragen @davidpgil This thread is fascinating, just commenting so I can follow along :)
(DIR) Post #298269 by uranther@cybre.space
2018-09-30T13:25:28Z
0 likes, 0 repeats
@kragen @vertigo @davidpgil @jjg I guess I was just referring to what "very far" means in this note."I should point out that RISC-V raw binary images share this characteristic; you can get very far indeed with nothing but raw binary images with this processor ISA. Better yet, you don’t even have segmentation-imposed alignment restrictions like you do in x86 real-mode!"So x86 raw binaries cannot appear just anywhere in real-mode address space?I am just trying to understand :blobcatcoffee:
(DIR) Post #298286 by uranther@cybre.space
2018-09-30T13:28:50Z
0 likes, 0 repeats
@vertigo @kragen @davidpgil @jjg If there isn't a modern executable format that's not ELF and that supports 64-bit PIEs... then what will you call your new one? :blobthinkingcool:
(DIR) Post #298332 by uranther@cybre.space
2018-09-30T13:35:49Z
0 likes, 0 repeats
@vertigo @kragen @davidpgil @jjg Pardon the omission. + @calvin
(DIR) Post #298374 by uranther@cybre.space
2018-09-30T13:42:45Z
0 likes, 0 repeats
@calvin XCOFF64 looks modern enough :thounking: @vertigo @kragen @davidpgil @jjg
(DIR) Post #300956 by vertigo@mastodon.social
2018-09-30T17:38:00Z
0 likes, 0 repeats
@uranther @kragen @davidpgil @jjg Most older instruction set architectures have instructions or addressing modes which references data in absolute terms -- that is, if I say something like MOV EAX, [4], then EAX will be loaded with whatever value is at address 4. Unless you're relying on segment or page-based relocation methods, that's the sole address you will fetch from.
(DIR) Post #301111 by vertigo@mastodon.social
2018-09-30T17:54:44Z
0 likes, 0 repeats
@uranther @kragen @davidpgil @jjg Honestly, I'm just going to use Hunk Format. There's no need to reinvent that specific wheel. It has existed since I was only 6 years old, and it's survived the 68K -> PowerPC transition under early AmigaOS 4.x, so I see no reason not to adopt it for my own use.It's main value will be in helping me to port an ELF loader (naturally; ELF has to be supported eventually) and to get a working VertigOS environment running sooner.
(DIR) Post #301127 by vertigo@mastodon.social
2018-09-30T17:57:14Z
0 likes, 0 repeats
@uranther @calvin @kragen @davidpgil @jjg I looked at *COFF formats, and I'm not impressed with any of them either. They serve as great examples of where the "lob everything into a common file format" thinking has gone wrong, in my opinion.I'll just use hunk format for my own tools. You can make a fully functional loader for it in less than 50 lines of C code. Smart data structure, dumb code is what I'm shooting for.
(DIR) Post #301536 by uranther@cybre.space
2018-09-30T18:32:08Z
0 likes, 0 repeats
@vertigo I read the rest of your second post.Fascinating stuff! And I've got more to learn. Thanks for sharing your knowledge and insight.@calvin @kragen @davidpgil @jjg
(DIR) Post #301595 by davidpgil@floss.social
2018-09-30T18:35:19Z
0 likes, 0 repeats
@uranther @vertigo @calvin @kragen @jjg Yep im just listening real carefully. great stuff!
(DIR) Post #A2mDDZQ13c8Kd5kSy8 by kragen@nerdculture.de
2018-09-29T23:09:51Z
0 likes, 0 repeats
@davidpgil (The other drawbacks of Z80s is that they're rather slow—several clocks per instruction, and only 8-bit ALU operations—and, despite what OriansJ said the other day on #bootstrappable, I suspect ARM Thumb beats them on code density. Aside from machine-efficiency issues, bank-switching is kind of a pain.)
(DIR) Post #A2mDDaWmvv924NvOAC by davidpgil@floss.social
2018-09-29T23:15:44Z
0 likes, 0 repeats
@kragen ive been focusing on the #6502 lately and am about to tinker with recreating a #NES game intro... taking it from there. Im sure all of this is slow but I think thats not the point. its about bringing quality/creative development back into computing, imo.
(DIR) Post #A2mDDbYF7zu1FBc44O by kragen@nerdculture.de
2018-09-29T23:29:13Z
0 likes, 0 repeats
@davidpgil Yeah, the 6502 certainly has a proven track record of people producing cool creative work after only a few weeks or months of learning. Part of that was being hooked up to special-purpose graphics and sound hardware that you sort of "scripted" in 6502 assembly, as https://prog21.dadgum.com/173.html explains. The #NES, as you probably already know, has similar hardware.Arduino and Processing have been pretty successful at making this kind of stuff accessible, too.
(DIR) Post #A2mDDblMLEOVtrkXVw by davidpgil@floss.social
2018-09-29T23:33:03Z
0 likes, 0 repeats
@kragen i am actually quite interested in learning low level #programming eapecially after having spent so much energy i being locked into coding ecosystems. all i care about right now is learning #assembly for various chipsets and moving into using what i know of #C to deal with more high level stuff. thanks for your link. ill check it out soon.
(DIR) Post #A2mDDcMa6qAVlJAGX2 by kragen@nerdculture.de
2018-09-29T23:35:20Z
0 likes, 0 repeats
@davidpgil I think #C is probably the ecosystem with the lowest lockin at this point, but anything that's purely free software is a reasonable candidate. Although I don't think anyone is going to implement a full DOM for the Z80 anytime soon, despite the existence of SpiderMonkey and V8 and so on.
(DIR) Post #A2mDDdDOwSQ4P7sRrU by kragen@nerdculture.de
2018-09-29T23:41:15Z
0 likes, 0 repeats
@davidpgil A big disadvantage of #assembly is that you're locked into a particular CPU architecture; #C helps with that. Although it's not clear what equivalent to C is going to bridge Vulkan, GLSL, C for your CPU, and Verilog for your FPGA—if anything.
(DIR) Post #A2mDDdbVUphZbt9hR2 by davidpgil@floss.social
2018-09-30T00:06:28Z
0 likes, 0 repeats
@kragen #LISP ? ... I know it sounds like a joke. I actually dont know much about LISP compared to #C ... I dont know that it emerged from #Maths and to me, by intuition, seema like we all should be using that. I just heard of #Forth today from the guy making his own #OS
(DIR) Post #A2mDDeOQYwpk3c2lge by kragen@nerdculture.de
2018-09-30T00:08:28Z
0 likes, 0 repeats
@davidpgil #Lisp definitely has its advantages; you may be interested in http://canonical.org/~kragen/sw/urscheme and https://github.com/davazp/eulex/blob/master/lisp/lisp.fs. I think #Lua is probably a better #bootstrapping choice for the Lisp level, though.
(DIR) Post #A2mDDedfeH1iotAwRk by davidpgil@floss.social
2018-09-30T00:11:35Z
0 likes, 0 repeats
@kragen I was almost talking about #Lua today regarding #Pico8 and how it seems to be a very nice place to prototype ideas for software (games) because everything is small and lmited by design. the equivalent of thumbnailing before u paint the whole portrait for real.
(DIR) Post #A2mDDf0MHvAtxFn3oG by feoh@cybre.space
2020-10-02T16:24:18Z
0 likes, 1 repeats
@davidpgil @kragen Totally agree! also goes for tic-80 which is fully open source!https://tic80.com/