Which OS would you recommend?

ralphbsz

Daemon

Thanks: 769
Messages: 1,299

#26
Actually, the mouse predates even Xerox PARC: It came from Doug Engelbart at SRI. I think the original mouse (which is patented) may be in the Computer History Museum in Mountain View today; I'll have to ask my friend who volunteers there.

It is well known that Apple stole nearly all of its good ideas, and many people correctly point out that much of that happened when Steve Jobs (who famously had no morals whatsoever) visited a demo of the Alto at Xerox. But if you look around, you find that Xerox also stole most of the ideas. Which shows the following: ideas are easy and free. The hard stuff is taking a bunch of good ideas, knowing which ideas are actually good and crucial and which ones only look good and are actually fluff, and turning it first into a functioning prototype (which PARC succeeded in). Even harder is turning a prototype or a demo into a commercially viable product (which Apple succeeded in). In spite of all the bad things we say about Apple, we have to give them credit for bringing a GUI to the masses.
 

Crivens

Moderator
Staff member
Moderator

Thanks: 688
Messages: 1,645

#27
Is that museum visitable online? I would like to see it, but since it is on the other side of the globe that is not so easy.
 

ralphbsz

Daemon

Thanks: 769
Messages: 1,299

#28
Not really visitable online. There is a lot of information at computerhistory.org, including a large photo gallery from their exhibits. But to really see an IBM 1401 or PDP-1 working you just have to be there.
 

Sensucht94

Well-Known Member

Thanks: 357
Messages: 374

#31
To add something to what lebarondemerde already said,
another interesting distro, if you prefer compiling from source, is CRUX. I second the choice of Ravenports for lightweightness, compatibility, cross-platform compiling purposes, even though, on the other hand, Void's xbps is extremely good on its onw already. Other good choices for multi-plartform portable packaging systems are pkgsrc, especially if using Slackware as base OS and nix. I have used Slackware for years before switching to Void and it's really a nice GNU/Linux distribution. Beware however that although I like distro-hopping and trying on Linux the software, the FSs and the inits I can't have on BSDs, eventually I always fall back to FreeBSD and NetBSD, which is where I ado all of my study and my serious everyday desktop computing and home server appliances; I'd use DragonflyBSD more if my machines weren't all equipped with Nvidia; Linux to me is more of leisure activity, mostly driven by curiosity and fun, but that's just an opinion of mine
 

Rigoletto

Daemon
Developer

Thanks: 773
Messages: 1,691

#33
RHEL you pay for support, including packages updates. But there is CentOS clone of it.

I do not know how are the contracts related to AIX but as far I know that just run on PowerPC hardware.
 

ronaldlees

Aspiring Daemon

Thanks: 300
Messages: 709

#37
Also, if you just want to learn then look at A2 Bluebottle which is the latest Oberon by Wirth. If Oberon and Plan 9 are too exotic then Minix is probably a better choice for learning “UNIX” because of its leanness.
I have a really old machine (circa 1999) that runs A2 Bluebottle. It works, but the browser puts me in mind of Mosaic. None of my newer machines work with A2, so it'd probably be an issue for most folks. However; the same people (Uni @Zurich) have created Barrelfish which may be more likely to run on recent hardware, and is compile-able for ARM platforms. Not very related to Unix.

Plan9 is about uber-networking, with a potentially much larger network profile (even for cpu sharing, etc). I have the gut feeling that it'd be more difficult to secure due to the larger network surface area, but I guess I have no anecdotal evidence for the feeling. As ralphbsz wrote, it's not the original Unix in any way.

Www.PDP11.org would give you an emulator for that ancient stuff, if you'd want to work your mind (but not much else :) ).

Minix3 is probably somewhat Unix inspired (after all, Tannenbaum did it - IIRC - because there wasn't a free Unix at the time).

The best choice that's doable? Go back to FreeBSD 3, which was very lean. It'll upgrade you from a Mosaic-like browser to an early Netscape one (quite an improvement over A2's browser IMO). AT&T forced BSD to change the code, but it's no doubt still spiritually connected to true UNIX. There won't be a canned image for your USB tho :-(

http://ftp-archive.freebsd.org/pub/FreeBSD-Archive/old-releases/i386/

BTW, welcome to the forum!
 

Spartrekus

Well-Known Member

Thanks: 52
Messages: 356

#39
Then go with FreeBSD. Although it may not be perfectly pure to some old version of UNIX its still a great darn-near-close-to-real-UNIX OS, and the support on this forum is top notch. Since its learning you principally want to do with it.
There is nothing interesting in Pure UNIX. there is only dd, ls, cp, ... and a basic C compiler. What can be interesting in that?
 

Spartrekus

Well-Known Member

Thanks: 52
Messages: 356

#42
You should check out pcc, the original portable C compiler. They keep it running on a pdp11 and the 2.2BSD folks are also around.
pcc might be in any cases better than tcc.

there is too a legacy cc (machine PDP):
https://github.com/mortdeus/legacy-cc

The original CC V7 is here: http://www.tuhs.org/cgi-bin/utree.pl?file=V7/usr/src/cmd/cc.c
You just need to compile the lib.
It (cc.c) is not a long code, isn't it? ;)

Look carefully this from the above, original cc.c code:
Code:
# include <stdio.h>
# include <ctype.h>
# include <signal.h>
Impressive today for a modern programmer ...
 

Crivens

Moderator
Staff member
Moderator

Thanks: 688
Messages: 1,645

#43
That cc.c is only the driver code, not the compiler complete. In those days, compilers were multi-passed and had a lot of stages.
 

Spartrekus

Well-Known Member

Thanks: 52
Messages: 356

#44
to something parent
That cc.c is only the driver code, not the compiler complete. In those days, compilers were multi-passed and had a lot of stages.
At this time, maybe you know why so many stages?
Denis R. wrote it, right? He must have spent many months/years working on it. He wrote really a lot (see above link with source / *.c comments).
 

kpedersen

Daemon

Thanks: 398
Messages: 1,245

#45
Ken Thompson's 8c from the Plan 9 OS is really lean. If I recall, it was also used by Google's Go compiler in the early days (pre 1.3 IIRC) to compile the Go code into native machine code. This means you can still obtain a port of it to FreeBSD and Windows if you grab it from the old Go archives.
 

Crivens

Moderator
Staff member
Moderator

Thanks: 688
Messages: 1,645

#46
At this time, maybe you know why so many stages?
The pdp11 had what? 64K of memory? There simply was no room to fit it all in.

Does gcc still have 3 separate programs (cpp, cc1, as)? That is one of the clang speedups, no separate passes for one compile run.

I know of one compiler that had 7 passes (tokenizer, parser, syntax check, semantic check, attribute generation, code selection, code generation) as separate passes. None was bigger than 16 KBytes and the intermediate format was files on disc. This was needed when you have 64K and need to have dynamic data as well as the OS in memory. Optimizer passes were optional.

Would you be able to do that today, you would also be able to multi-thread one compile run. When one C++ file takes minutes to compile, that may be beneficial.
 

Spartrekus

Well-Known Member

Thanks: 52
Messages: 356

#48
The pdp11 had what? 64K of memory? There simply was no room to fit it all in.

Does gcc still have 3 separate programs (cpp, cc1, as)? That is one of the clang speedups, no separate passes for one compile run.

I know of one compiler that had 7 passes (tokenizer, parser, syntax check, semantic check, attribute generation, code selection, code generation) as separate passes. None was bigger than 16 KBytes and the intermediate format was files on disc. This was needed when you have 64K and need to have dynamic data as well as the OS in memory. Optimizer passes were optional.

Would you be able to do that today, you would also be able to multi-thread one compile run. When one C++ file takes minutes to compile, that may be beneficial.
(Except respected Ken). Do you think that today a single programmer would be capable to realize a C compiler just from scratch?
- I am not so sure if this guy today still exist.
 

Crivens

Moderator
Staff member
Moderator

Thanks: 688
Messages: 1,645

#49
One C compiler? Sure. Competive production quality? No. I once worked with a guy who had written a complete C++ compiler all by himself. In assembly. Mindboggeling blindingly fast, code quality was, well, on par with single pass compilers. And that was not his first compiler to be written in assembly. You need to be half aspie to talk to such people and be taken serious as well as being able to follow them.

Ah, that were the times.
 

Spartrekus

Well-Known Member

Thanks: 52
Messages: 356

#50
One C compiler? Sure. Competive production quality? No. I once worked with a guy who had written a complete C++ compiler all by himself. In assembly. Mindboggeling blindingly fast, code quality was, well, on par with single pass compilers. And that was not his first compiler to be written in assembly. You need to be half aspie to talk to such people and be taken serious as well as being able to follow them.

Ah, that were the times.
"ah that were times". you mean that it was good time. But, you mean that such programming C/C++ compiler is no longer of big impact or any interests today?


--
Listening nradio (ncurses)?
https://github.com/spartrekus/nradiofm
 
Top