Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Comparison between Collapse OS and CP/M #52

Open
hsoft opened this issue Oct 16, 2019 · 28 comments

Comments

@hsoft
Copy link
Owner

@hsoft hsoft commented Oct 16, 2019

A lot of comments I get with regards to Collapse OS is "Why not CP/M?", which is a very fair question. I need to come clean with this answer: I haven't tried very hard yet to have CP/M self-host on a minimal machine. This needs some context.

When I first had the idea of Collapse OS, I wasn't very knowledgable about OSes (I still am not). I had no experience with Z80, my first dreamed up designs were with AVR MCUs. As I surfed around for prior art, I found Knight OS, which I found really cool and made me choose Z80.

Then , I digged some more into the Z80 universe and, of course, found CP/M (and others, such as FUZIX). My first concern was self-hosting. I knew Knight OS was not self-hosted, I asked @EtchedPixels whether he thought FUZIX could eventually self-host on its target machine and he thought that it wasn't reasonable to work towards this. I did attempt self-hosting on CP/M, but failed. I didn't know whether it was even possible and couldn't find an easy answer online. I abandoned and started Collapse OS as we have it now.

I'm a "hands on" person. When I don't know about a subject, I need to get my hands dirty, which I did. Now, this project went viral beyond my wildest expectations and here we are.

Now, as @EtchedPixels is telling us in #49 (thanks for the insight!), CP/M is self-hosting, which is a new information to me.

This issue's goal is to gather around information about CP/M's self-hosting capabilities, compare it with Collapse OS' goal and see if continued effort on Collapse OS is worth it or if we should rather pour effort in David Given's CP/Mish

Now, for the record, Collapse OS has some really impressive self-hosting capabilities: it self-assembles with less than 8K of RAM and less than 16K of ROM. Of course, I really doubt that CP/M can do that, but it's still interesting to compare capabilities and assess tradeoffs.

So, does someone have more information?

EDIT: this message originally referenced BDS C but I had my memory mixed up. CP/M from Digital Research is ASM code. A C compiler isn't necessary to assemble it.

@hsoft

This comment has been minimized.

Copy link
Owner Author

@hsoft hsoft commented Oct 16, 2019

In fact, I think I'll afford myself a ping to @davidgiven (if you don't mind...)

@f4grx

This comment has been minimized.

Copy link

@f4grx f4grx commented Oct 16, 2019

@jes

This comment has been minimized.

Copy link
Contributor

@jes jes commented Oct 16, 2019

From a skim of the cpmish README, the project sounds cool but it's not yet self-hosting.

You pretty much need a Unix --- I developed it on Linux. You'll need to install the dependencies. These are the names of the Debian packages: [...]

@hsoft

This comment has been minimized.

Copy link
Owner Author

@hsoft hsoft commented Oct 16, 2019

@jes yes, I understand that, but considering licensing issues associated with code from Digital Research, working on CP/Mish could be, I think, the best option to provide an open source project achieving Collapse OS' goal within its defined constraints.

@keithstellyes

This comment has been minimized.

Copy link
Contributor

@keithstellyes keithstellyes commented Oct 16, 2019

Honestly to come clean, when I first saw CollapseOS, I was honestly thinking "Why not just use CP/M or SymbOS?", but there is no issues of copyrights, licensing, etc. here. Plus I just wanted an excuse to program myself some Z80 OS code.

In fact, I never did get fully convinced, but I think the big-ticket item for me is modularity that sold me on CollapseOS, CP/M isn't running on TI-84s, and it's fascinating to see a simple OS kernel that can in theory be easily ported to all these different systems

@jes

This comment has been minimized.

Copy link
Contributor

@jes jes commented Oct 16, 2019

Another reason to prefer CollapseOS over CP/M is that CollapseOS can run without any bank switching, whereas CP/M requires the bottom of memory to be writable.

https://retrocomputing.stackexchange.com/questions/6442/why-did-cp-m-require-ram-in-the-bottom-part-of-the-address-space

(OK, even with CP/M, you could put your ROM at the top of memory and then manually toggle in a jump to the ROM using the front panel... if you have one).

On a scavenged system you're likely to have neither a front panel nor bank switching.

@EtchedPixels

This comment has been minimized.

Copy link

@EtchedPixels EtchedPixels commented Oct 16, 2019

You don't need banking - just switchable ROM which is easy and you'll find people routinely retrofitted to old machines. On a scavenged system you won't find a Z80 these days either! Switched ROM is trivial. And btw you can recompile your CP/M and apps to use a different base. Tandy did that.

CP/M is sort of BSDish probably and originally in PL/M, but usually these days nobody bothers with the PL/M. There are open CP/M equivalents like the Z System which are in asm and is also self hosting (it had to be because that was what people had at the time!)

CP/M needs more than 8K RAM - a fair chunk more but almost no ROM. However nobody has made RAM that small for decades.

Really though it's not a technical question - are we having fun ? is the question 8)

I'd have to try reassembling the Z system with Zmac on different memory sizes as all my CP/M boxes (like pretty much any CP/M box once 64Kx1 RAM came out in 1982/3 or so) has 64K or more.

@davidgiven

This comment has been minimized.

Copy link

@davidgiven davidgiven commented Oct 16, 2019

cpmish isn't self hosting for two reasons: firstly, while the CCP, BDOS and BIOS are all in classic-syntax asm which can be compiled with the tools on CP/M itself, it's painfully slow to do so on a real machine; and secondly, some of the tooling is in ANSI C which doesn't really self-host on CP/M. There's only one native ANSI C compiler for CP/M, Hitech C, but it's very very slow and the syntax is kinda weird. Frankly, fitting a C compiler into CP/M is hard.

I do have a technically self-hosting Adaish compiler which runs on CP/M (http://cowlark.com/cowgol/) but it's way too slow for any real use. I'm currently reworking it to be much smaller and simpler --- see http://cowlark.com/2019-07-15-cowgol-prototype if you're interested. I'm hoping this one will work better on real machines.

@earney

This comment has been minimized.

Copy link

@earney earney commented Oct 18, 2019

First I have to admit I know nothing about CP/M, but how about forking it, and stripping it down (removing unnecessary code/features that violates the mission of this project)? Then add the features that are missing. It seems that CP/M has quite a bit of history, and a lot of tools already developed. From a few videos I've watched it is an ancestor of DOS, which a lot of people are somewhat familiar with.

I know that creating a new OS would be cool (and could be a side project), but a CP/M ish type OS would be more beneficial in a collapse event more than a new OS.

@davidgiven

This comment has been minimized.

Copy link

@davidgiven davidgiven commented Oct 18, 2019

There are no unnecessary features in CP/M. It is literally the minimum functioning disk-based operating system it is possible to have.

If anything it needs more features: one of its crippling drawbacks is that there's no generic means of drawing text on the screen (because it predates common glass TTYs), requiring all programs to essentially contain their own curses equivalent. It desperately needs an extra BDOS/BIOS call for screen manipulation.

@boomlinde

This comment has been minimized.

Copy link

@boomlinde boomlinde commented Oct 31, 2019

@davidgiven Correct me if I'm wrong, but a CP/M application could theoretically load screen manipulation routines with a standardized interface from a disk at run-time as necessary. These don't strictly need to be part of the BIOS for there not to be re-implemented on a per-program basis. Existing applications for CP/M would either way not gain anything from it, in which case I think that whatever results in a smaller kernel/BIOS is preferable.

My own gripe with CP/M is that it's inherently disk based. It's designed around the idea that you have some kind of seekable block device. It makes it unsuitable for simpler linear storage media like magnetic tape.

@davidgiven

This comment has been minimized.

Copy link

@davidgiven davidgiven commented Oct 31, 2019

They could load the screen routines at runtime, but I've yet to see a program which did. High-end apps have complex general purpose screen code which is then patched by a configuration program with specifics about the terminal --- so you always carry the cost of having, e.g., code for supporting line insertion, even if your terminal doesn't have it. It's a complete nightmare.

I'd much rather have generic BIOS entrypoints for screen drawing and then lose the expensive terminal emulation code from the BIOS; without the terminal emulator state machine you end up with drastically simpler and smaller screen handling code throughout. If you really need a terminal emulator, load that on demand.

(Extending CP/M with the ability to load TSRs would be incredibly useful, too.)

Also, re disk support... I have an Epson PX-8 CP/M laptop, which runs off tape. Please don't ask if it runs well.

@EtchedPixels

This comment has been minimized.

Copy link

@EtchedPixels EtchedPixels commented Nov 1, 2019

@boomlinde it could - in fact the MSDOS world fixed it exactly that way with a BIOS that did cover basic graphic support and a model that allowed overriding by ROM or software. (The BIOS all being in ROM is a later IBMism)

A tape based OS is fundamentally different to a block storage one in just about every way imaginable.

@Allisontheolder

This comment has been minimized.

Copy link

@Allisontheolder Allisontheolder commented Nov 11, 2019

I've used CP/M since the 70s and when P2DOS a z80 clone was introduced I used it.
FYI it was assembled using CP/M as the development environment with no issues for
space to re-create. In fact I've use cp/m systems to assemble CCP/BDOS using native
tools. After all it was always done that way. FYI Ron Cains Small C needed 48K to
compile itself and maybe 200K of disk space.

The minimum functioning disk system was Northstar dos, it fit in about 0A00h bytes
that 2560 bytes. Enough to load, save, and run programs. anything else was loaded
as a utility.

CP/M is not overstuffed. Its a complete file system with scatter/gather media usage
and a complete set of utilities and tools with it but stored on the media so if not needed
delete them. A typical system build is about 8K usually it can be a bit less. When a
program is loaded the CCP the use interface is overlaid and on return its restored.
Same for the BDOS is more space is needed and yes on return it is also restored.
The BIOS is in ram (or Eprom if one chooses) and nominally resident as its all of the IO.
So a system can have a memory space with BDOS resident of about 4K to 5K less
than 64K. Typical was 56-58K. With it brings a large library of accessible programs.
(Walnut Creek CP/M CD is on the 'net and has 85% of everything ever). Its real advantage
is anything block replaceable for storage is possible and has been done. That covers
floppies (all sizes), MFM and RLL hard disks, SCSI disks, CDrom (scsi), and IDE,
CF, SD, uSD, EEPROM and FLASH, and tape!

As to a compact OS for minimal hardware likely most here have no knowledge of system-1
or the similar integrated packages that would runn on 8080/8085/z80. This is a 1976 this that
ran on the first Altairs and IMSAIs with maybe 8k of ram and was loaded from paper tape of audio cassette. An example of this that can be fond on the 'net is Processor Technology ALS-8 and it contains a monitor (as called here a shell), and assembler and simple line editor
(CRTs were later!). It was enough to given about 10K of ram to assemble itself in modules.
ALS-8 is romable so it was also available as a board with Eproms as well TEH PT SOL
was among one of the first complete integrate terminal (CRT, Keyboard, and 8080 CPU)
systems.

RE: PX-8 the tape was terminally slow but worked. The other approach they took
a wedge that was an attachable ramdisk of 120Kb with battery backup. I have
two they work very well though every ten years the NiCds need to be replaced
(they are ca 1984ish new). There was also a external Floppy with about 350K
(5.25 and 3.5 inch). For those trying to grok programming in 120k , it is adequate
since CP/M is in rom and same for utilities and Basic. I have three, two with 120K
wedge and one with 64K wedge.

Most modern OSs for small system are written in C and invariably the OS and supporting
east something int he 32K region even UZI-unix. That is not so bad but there is not a large
catalog of useful programs. Those being Editors, assembler, loaders, spreadsheets, database
tools.

Tape based OS can be a block replaceable system and the only difference is speed
for access. I have a few that still work such as DEC tu58. Those are 256K of storage
per DC100 (two drives per deck) tape and about 60 seconds for a full rewind and
search for first. Some however are serial ( the whole file runs as several/many
blocks) file systems and are compact but slow, still useful.

In the dark ages (1970s to early80s) tape could also be the common then portable
cassette recorder and the data was encoded as tones for writing and read back as
if it was paper tape. Cheap as in minimal electronics and those recorders were
typically 40-50 bucks then, simple and remarkably effective. Usig two you could
read from one ans write to the other as a crude database system or assemble
and write to tape.

Of course if one is into TEOWAKI [the end of the world as we know it] then I suggest the
1802 CMOS CPU as its rad hard compared to NMOS and if you have to build it from
transistors you will use a whole lot less of them.

All one has to do is look back for the answer instead of reinventing the incompatible wheel.
Enough for now.

Allison

@Allisontheolder

This comment has been minimized.

Copy link

@Allisontheolder Allisontheolder commented Nov 11, 2019

and an OBTW...

CPM is not assembled as a monolithic lump. You can but really?
If you are its likely from an in memory disasembly.

If you are then your knowledge of CP/M is faulty or at best incomplete.
No excuse for that as the needed DOCS are on the 'net.

The assembly process for one of those like by Clark A. Calkins 1981 disassmbly
which is about 96K bytes total. However that is the hard way. First create a
header file from that, then extract the CCP module, BDOS module, and then your
own CBIOS. They are assembled with the header file added in and then merged in
memory and saved as needed for later use. And yes you can use ASM unless
the disassembly is not asm compatible (they exist) then at that point you use what
they used and likely easily found on the net. I like ZASM, XASM80, and MAC80.

Learn marco assembler they can make life easier. There are more than a few
available for CP/M. A linker is handy for modular builds. Though LOAD and DDT
or ZLOAD and ZDDT work too. The Zxxx is a z80 coded version, or Z80 specific,
usually.

The largest lump in that case is 71K of text and code, delete the comments
and its get much smaller. The assembled code for that would be a bdos which
assemble to 3.5K.

If I can do that back when in a small system, 40K total of ram so can you.

My first BIOS was build using PT ALS-8. That was used to get the system
binary off a 8" SSSD disk that was standard back when. Then once loaded in
memory used to run a program called MOVCPM. Early days you never had
the source but the base 20K binary was on disk and a tool to move it to the
desired address and you then liked you CBIOS to it and did a PUTSYS to
write it back to the disk. Right around 1979 the reconstructed source (using
RESOURCE, later ZRSOURCE) made it possible to do that from a source file.

Size of things:
Assembled CCP is 2K of 8080 code. Same for ZCCp and clones.
Asembled BDOS will always be 3.5K even ZRDOS, ZSdos, novados, P2DOS
Asesembled BIOS is variable depending on user hardware and user defined features added.

Why are those sizes fixed: CP/M had standards and the clones even if improved
adhered to that so any of those modules assembled for the correct address would
drop in. Its features of the day was hardware, as in your box and disk, were
abstracted from the OS and most software. That allows the base system and
user software to run equally well (or faster) on 8080, 8085, z80, z180, z280
with a suitable bios.

One last trick... the system does not have to load from disk, the boot program
does that. It can copy it from a Eprom (8k is big enough) to ram and jump to it.
So for my 1979 CP/M 2.2 system four 2716s made booting lighting quick
The bios for that new how to access rom (in IO port space) and copy as
need to ram. IT was the first system I had with BIOS and a monitor in eprom
and it made life it easier as system was improved with more modern disks.

PC dos copies much of that as their starting point and no it was not a translation
from 8080 to 8086. There is a CP/M 86 and also a version that ran on the Z8000.

Much of the premise of CP/M can't assemble itself is based on error or misinformation.

One last thing, the Unofficial CP/M Repository was how the then (before Lineo)
owner gave permission to Tim to post that stuff without their direct involvement.
Time then worked with Gaby to get it a permanent home back before 2000.
I helped Tim compile a lot of the early stuff like the disassembled sources.
FYI they never said you could not give it with something only that its availability
must be cost free. so it was a way to get the xyz computer and then download it.

Allison

@hsoft

This comment has been minimized.

Copy link
Owner Author

@hsoft hsoft commented Nov 11, 2019

@Allisontheolder you seem emotionally affected by a claim that isn't made. It is not claimed, in this ticket, that CP/M isn't self-hosting. I say so in the first message.

That my knowledge on the subject is faulty is evident. I say so in the first message.

Thanks for your juicy details, they're not devoid of interest, but they don't help much advancing the subject at hand.

@hsoft

This comment has been minimized.

Copy link
Owner Author

@hsoft hsoft commented Nov 11, 2019

OBTW: regardless of anything an author say or don't say, default copyright law apply, so if code isn't properly licensed, pouring work into this code and redistributing it is a legal landmine.

@davidgiven

This comment has been minimized.

Copy link

@davidgiven davidgiven commented Nov 11, 2019

Be careful of the Digital Research stuff on the Unoffical CP/M site --- the license allows it to only be distributed on the Unofficial CP/M site. i.e., if you copy it and use it elsewhere, you're breaching the license. It sounds crazy but that's what it says and I confirmed that interpretation with Gaby Chaudry. This is why I put together cpmish, which is verifiably (or at least plausibly, for some of it) properly open source. (Sadly I was unable to find a free BDOS and CCP clone for the 8080, which is why it'll run on a Z80 only.) (A lot of software back from then has really, really dodgy licensing. The overwhelming majority of stuff I've found on PD archives is unusable.)

@Allisontheolder

This comment has been minimized.

Copy link

@Allisontheolder Allisontheolder commented Nov 12, 2019

All at once,

First do not email me without prior permission. Helps keep you from being dumped to spam.

The comment about emotion is ad hominem , we can do better.
If we didn't have passion for this stuff would we bother? Not
like there's is money in it.

Second I have my own license actually three, Lifeboat, Northstar, and the usual one
with every SSSD 8" copy purchased.

How many here claim that the BIOS is part of CP/M and cannot under license
be altered and distributed? News flash its not the copyright holder for the BIOS
(only a specific one that's used only for the SSSD 8" copy as a working example)
is varied I hold one for my CBIOS, Nothstar, Compupro, and Lifeboat to name
a few are also for their specific versions. There are many others.

To repeat CP/M is a modular system with CCP, BDOS, Cbios(user or manufacturer
supplied) and all the utilities (ED, pip, load, ddt, asm, and a few more...).
IT has no impact if I substitute equivilents such as:

ZCCP, P2DOS, CBIOS and utilities (PIPX, loadX, as so on) as its been done.
Any license there is not to DRI but the individual holders that put them in the
PD for everyone to use nearly all under GPL or copyleft, a few just have
keep my name on it.

As to distribution as it stands If I hand you a copy on a disk whats the
problem. That's assuming I've not altered any attribution of copyright
statements? that includes the release by Lineo.

Someone comments that CP/M due to that is not worth investing time as
you can not alter it. First your do not have to and the CBIOS is your business.
Please read the enclosed license from the site by bryan sparks of Lineo.
Note we are the "Community". Tim Olmstead worked hard to get that.

Here is the license:

Subject: Re: Unofficial CP/M Website/licensing of CP/M material
To: gaby@gaby.de
Date sent: Fri, 19 Oct 2001 10:36:31 -0600

Let this email represent a right to use, distribute, modify, enhance and
otherwise make available in a nonexclusive manner the CP/M technology as
part of the "Unofficial CP/M Web Site" with its maintainers, developers and
community.

I further state that as Chairman and CEO of Lineo, Inc. that I have the
right to do offer such a license.

Lineo and its affiliates, partners and employees make no warranties of any
kind with regards to this technology and its usefulness or lack thereof.


Bryan Sparks
CEO Lineo, Inc.
http://www.lineo.com

Earlier I saw a comment about bank switching, another fallacy.
CP/M2 is a single map OS aka flat memory no bank
switching is directly supported, that mind if you implement
something in the CIOS its your ball. CP/M+ (AKA CPMV3)
supports bands as will MPM.

And yes you need ram at low address as 100h is the
default load address for everything. Its no big deal to
support rom/eprom at high address and it was the
normal thing.

Now why here...?
This is only the second OS I've found on the 'net that is assembler
only and for older 8bitters (Z80). That is significant, Retrolium is
the other. That and I've been building my own just because after
40+ years of 8080/8085/z80 I cant break the habit.

So I can give you a disk with a CBIOS and nominal CP/M and
as long as its labeled as copyright DRI/CDR/Lineo its just fine.
Remember I didn't sell it or anything and the copyright at
Unofficial CP/M site is for personal use only.

Anyhow your likely not listening by now, so back to our sponsor
and keep at it. I can not contribute code as it may violate
the clean room rules. It doesn't stop me from looking at
it, assembling it and testing pieces.

FYI This account new only because the others present issues with
software I have worked with. I'm an old long time user of GIT and
other SVNs.

Allison

@davidgiven

This comment has been minimized.

Copy link

@davidgiven davidgiven commented Nov 12, 2019

ZCCP, P2DOS, CBIOS and utilities (PIPX, loadX, as so on) as its been done.
Any license there is not to DRI but the individual holders that put them in the
PD for everyone to use nearly all under GPL or copyleft, a few just have
keep my name on it.

Sadly, most of the stuff I've seen frequently does not have a license at all, which means I can't use it, or else has an unworkable license with clauses like 'for non-commercial use only', which means it cannot be safely distributed with another work. I've been putting together http://cowlark.com/cpmish, which is a provably open source CP/M (more or less) distribution with a proper paper trail. The intention is to have something free enough to distribute as a bootable image with emulators in something like Debian; Debian DVDs are sold for profit, which counts as commercial use...

Since 1977 (the Berne Convention), any work which does not have an explicit license is automatically copyrighted, all rights reserved, and is not distributable. I had enormous trouble finding usable BDOS and CCP versions because of this ---- I ended up having to go with ZCPR1 (public domain, AFAICT) and ZSDOS (GPL2).

You've been writing CP/M software for years --- please, I beg you, please put up a statement somewhere saying that you're making your work available under the terms of the MIT/BSD/GPL license. Otherwise there's a strong risk of it being lost forever!

Let this email represent a right to use, distribute, modify, enhance and
otherwise make available in a nonexclusive manner the CP/M technology as
part of the "Unofficial CP/M Web Site" with its maintainers, developers and
community.

'Its ... community', i.e. the community of the UCWS. Sorry. I double checked with Gaby Chaudry, and I'm not going to second-guess her. It means that the Digital Research software cannot be distributed via other channels without risk of lawyerbombs.

@EtchedPixels

This comment has been minimized.

Copy link

@EtchedPixels EtchedPixels commented Nov 12, 2019

The phrase 'non commercial use' is itself is also not useful. There is no legal definition of non-commercial use in quite a few major legal systems. It's a problem Creative Commons did a study on. It's also a 'magic' phrase - it means different things to different people and nobody agrees.

@Allisontheolder

This comment has been minimized.

Copy link

@Allisontheolder Allisontheolder commented Nov 12, 2019

HSOFT:
Copyright law in USA may differ from where you may be. There no copyright on this email
or any code I may post to publicly viewable forum unless...

I go >All work is property of (me) and is available for personal viewing only and is not permitted for use in whole or part without prior permission.

However if I state that everything written may be used or will not be litigated unless defamitory
that sets new rules.

I invoke the library rule. You can put my stuff on any bookshelf(web page) anywhere
but you cant sell it. Everyone then can read it and copy sections of it then.

In the end what Gaby said and repeated by a third part is mostly useless here.

Until someone with ATL or ESQ contacts me directly its all pointless and the legal
representative of Lineo has made it clear that the "owner" has no interest in litigating
there is no issue. Unless I try to make money off it.

Allison

@hsoft

This comment has been minimized.

Copy link
Owner Author

@hsoft hsoft commented Nov 12, 2019

Let's not burden this ticket with sterile bickering about open source licensing. @Allisontheolder, to borrow your own word (I hope you didn't copyright them), your knowledge on this matter is faulty or at best incomplete. Let's please stop there.

@Allisontheolder

This comment has been minimized.

Copy link

@Allisontheolder Allisontheolder commented Nov 12, 2019

Davidgiven:

Its all noise and means nothing. If there is a take down notice its easy to comply and same for cease and desist. To date (well over 30years) I've not heard of any. Layers will only work if they are being paid that means there is a financial interest or money to be had. Further some of the CPMish work may be questionable if your right, not withstanding with a dozen prior projects dating back to the days of comp.os.cpm and alt.sys.cpm (bitnet usenet and such).

So enough of so and so said and legal guessing, at least I have a IP law person to consult.

Allison

@davidgiven

This comment has been minimized.

Copy link

@davidgiven davidgiven commented Nov 12, 2019

@Allisontheolder That's your choice, of course, but there's also a risk. It's small, but present. These days third-party distributors such as Debian do license auditing specifically to avoid this risk, and this is exactly the kind of thing they ask about. No guessing is involved.

As I said, if you have any old CP/M software which you wrote that you'd like to formally release under an open source license, I'd be delighted to help you --- contact me off list. (I'm trying to collect this stuff.)

@Allisontheolder

This comment has been minimized.

Copy link

@Allisontheolder Allisontheolder commented Nov 12, 2019

Davidgiven:
Here in the USA we look at that statement as:
Nice place you have, shame if it brunt down. I can sell your insurance.
A threat thinly veiled.

Reality is nothing will happen. As to debian I've been using it since that fork was
created and Slackware before it (I started with linux back at kernal .9) never every
had a license or even a license check. Its a fully open and free system.

Anything offered up was into the usenet groups or on existing sites including Gaby's
as I helped Tim O. collect it (and contributed) before transfer to Gaby and in the 1990s.
The only losses were the one off groups that either lost their DNS or a bit of the Geocites
debacle.

That dispense with....hopfully.

@Allisontheolder

This comment has been minimized.

Copy link

@Allisontheolder Allisontheolder commented Nov 12, 2019

I'd offer at the present form, CP/M is not a comparison. CP/M was pretty warty.
TRSdos (TRS80) was a standard as well they sold hundreds of thousands of them.

Applications are the defining item not the OS itself.

I'd look to more primitive such as Northstar dos, or system-1 (ALS-8 or others such as
Altair software packge 1 and II) as they are generally enough to develop code and even run programs. They are all however space efficient and compact code.

There are two schools:
I want more than CP/M, your used to a PC with a gui or a sophisticated command line and want all that.

The other is minimalist, give me enough to get to the next level.

The second is where you started on 1975 with a bare Altair maybe even if your were
lucky a TTY (ASCII or baudot-5level code). 1K of ram then was 2 weeks pay after
obligations or about 199$US for 1K, 4K was 399$US. Software package II was
75$US, 45 for people that purchased Software package 1 (junk). Two years later
disk (5.25 with controller) was 599$US. I stated that as that was my development
environment at that time. No PC as that was marketed in 1981 and at 3500$ bare
it was pretty useless.

The difference now from then I if i were back then with knowledge possessed
now I'd have made greater effort for better storage and software tools. Also
I'd know what ones were likely better.

Now with PC and huge storage I'd start with a known entity and build a complete
disti (use a linux term) that I could easily launch on any 8080/8085 or Z80. That
would include basic tools to get it onto the bare metal without a PC. It would
not have to be CP/M. Current work is a Eprom with ALS-8, as that runs on a
minimal machine as a bootstrap environment. Then an OS with supporting
applications. Cromix would be great but applicaitons for it are non-existant.
Northstar dos and Helios are ok but applications are again thin. CP/M, lots of
applications and covering all the basic tools and even useful things like
spreadsheets (multiplan) and lots of editors (VDO and VEDIT), and
others. Oh right you need a terminal for that if your systems didn't
make it then all the old VT100s and the like didn't either. There may
be a old CRT TV to make a terminal with, maybe.

In the end if you had to fall back to 8080/z80 or 6502 what do you need?

A whole lot of stuff. All getting scarcer.

And networking is the first thing not available. Unless we go back to BBS
with nightly copy by modem transfers to other BBS systems.

Software is less than half when hardware is the platform.
How many can and have built a system without a premade PCB
and a bag of parts? Likely few.
How many can pull parts from a pile of stuff and make a
simple two transistor transmitter and a crude receiver?

Computers running software are an order of magnitude harder.

Allison

@hsoft

This comment has been minimized.

Copy link
Owner Author

@hsoft hsoft commented Nov 12, 2019

Please @Allisontheolder stay on topic. What you write (here and on other issues) is interesting, but often off-topic. The subject of pondering post-collapse conditions for bootstraping a computing platform is a very interesting subject, a crucial one even, but not the subject of this issue. Perhaps opening a new issue would be more appropriate.

Also, be aware that unless explicitly muted, all previous pariticipants to this issue receive a github notification on every comment. Off-topics comments like those can be seen as spam by some of them.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
9 participants
You can’t perform that action at this time.