You will find the PC/DOS thing really good. Most of its utilities have been defanged, so they run under many different DOS environments. In the fix pack, for DOS 70, you can make its true version appear to be 6.30: handy if you want to run QEMM.
But if all you are after is REXX, it's better to check out Quercus REXX as their rexx multi-tasks better. It also is available as a Windows version, where you can experiment with the bugs of the Windows command line: eg passing parameters to a script, eg "REXX frog a b" does not work.
OS/2 is strong where Windows is weak, and weak where Windows is strong. It's an interesting contrast. The OS/2 community, like the Linux community, attract people who want to be there, and the latest version out this year is not dissimilar to a Linux distro for what you get (except the price tag is higher). The next version is supporting a VMWare-like project that runs Windows and Linux in a virtual computer.
The trouble that I find with MS scripting languages is that they're unusually hard and restrictive. VBA for example, makes Excel calls by a rather lengthy call, and it's not much use for the causal programer (compared to say, 1-2-3's menu-keystroke command. [Yes, I used both].
So while I try to avoid MS stuff, just seeing how the other guys do it is an eyeopener. You would know yourself, that from your unix/linux experience, you make Windows work harder, and from your Windows experience, you make Linux work harder.
Gee. it's not only that. Everything in Windows is so deeply threaded that it is possible for a virus to lodge itself in the start-up sequence anywhere, and go unfindable.
For example, there are seven or eight differnt start-up objects in Windows 9x:
msdos.sys [hidden file]
config.sys
autoexec.bat
registry [many different keys]
system.ini
%windir%\system\vmm\*.* [just sucked up whole]
startup folders [yes, you can have startup folders nested.
What a program is to do with a file is done in three different ways as well.
It's little wonder that the thing is open to attack. You can't hunt it down unless you pretty much hack it, and follow their goofy retro thing with the 64-bit sequence: {01.22.23....}
I was thinking more in line with the/# command in the shell script: this passes the rest off to some other command, I think.
EXTPROC works like this. Suppose you have a batch ADD.CMD that goes like this:
EXTPROC RPN
b %1
b %2
a
s
The command ADD 1 2 would execute the command RPN ADD 1 2, which would read each line of ADD.CMD through RPN, and give the result of 5.
That's how I read it.
Whether you let REXX or CEnvi or Perl process the RPN script is a mute point. What I was getting at was that you could create commands for different processors, AND be heard correctly.
For recollection, Linux shells do support it. And yes, I have object and regina rexx for Win/OS2/Linux:)
The point I was making is that in Basic and Assembler, you go to a particular address, not a named function. Had I written 303 377, you may not had appreciated the point.
In the end, with the Basic RPN calculator, I used magic numbers to control the flow. This allowed for fewer labels, and robust code. The magic cookies were used to control the fan-out, with pre-testing of values (eg divide by 0) before it was attempted.
When the subroutine was done, it passed back magic cookies to indicate error codes, and its result in PA(0). It was up to the display module to announce the error message and deal with PA(0).
The message stack was also used for other things as well: you could peek at any register of the calculator.
But whatever it was, the codes were laid out so that all functions that would crash on DIV0 were together, all functions that required the range -1 to +1 were together, and so forth.
For the record, I used six magic cookies, in two sets of three. In half the cycle, set A would be controlling the flow, and set B were scratch data, while the other half B controls the flow, and A is the scratch data.
This greatly reduced the number of GOTO and GOSUB lines. I think in the end, there were 10 different labels of major jumps.
As for tweaking code: I still do it to great effect. I can still pull 8x increase in speed by selecting the correct algorithms.
When most people think of "computer", it's the desktop box, not some dinosaur in its air-conditioned pen with a side order of halon and a number of white-coats preening it. [Jargonfile quote] This is what I meant
Now, with Beowulf clusters, Distributed computing, and IBM runing Linux in mainframe VM machines, one is seeing lots of little PCs competing on mainframe turf. PCs have largely displaced mainframe terminals. A pc running a 3270 emulator is cheaper than a 3270 terminal!
At least I have the courage to not hide behind the "Anomynous Coward" monicker.
Still, of the four you quote, one bemoans the change of data and information. Another just agrees with what someone else's correction, a third is a hi to some troll, and the fourth is the story. This makes five.
I can find a lot to do in the evenings, thank you. It's just morning over here: nearly lunch time, actually. And yes, I am currently bored, apart from passing on my experiences in the old times.
I'd be more likely to while the hours away on Civ3 or REXX than read Ann Rice novels.
At least the OS/2 user that you trolled has written computer languages.
In OS/2 it is easy to do that. This is because of EXTPROC and REXX. From that, you can write a REXX script for the processor, and feed the language script in as data. Saved many hundreds of hours' work doing it like that.
Maybe you can do that in Linux too - don't know enough about that.
Most good authorities on computer languages says you should learn several languages. This allows you to think more creatively. I know.
that, when the mainframes ruled, that computing was associated with DATA (ie bits, bytes, fields and records), as in Automatic Data Processing, Datamation &c, but now that data is easy to come by, it's INFORMATION (eg Information Technology).
I wonder how many IT people suggest technologies that are not computer-related: eg how many people suggest paper cards as a solution. I know I have.
You see, once you start fiddling around with the hardware like Betty H did, you start using it wisely. It is one of the reasons that Unix works so well.
Before the computer revolution, computers were expensive and frail.
My computer at college in 1981 was something nearing the end of its life. It was an 8086 with 4K of ram, and a paper tape drive. To boot it, you load up the tape, and load three values into ram (by a series of eight switches and a "set" switch), and then send a command 377 to the processor. This would jump it tot a location in memory, and then run the commands that you loaded there (effectively JMP address), which would then run the KEX program. KEX was a driver for a teletype. After that, you input through the keyboard by assembler code.
Compared to that, mnenomics like a for add and b for bring would have been a godsend.
Of fortran, basic and cobol. In the days of wire wound core, each bit of the byte made the machine more expensive, and there was some comprimise on the size of the bit. Fortran was designed to run on a six-bit machine. Even Knuth's MIX is underpowered by modern computers.
BASIC is intended to run in small memory. MS made their packet by bumming it into 4K of ram, with a point and shoot interface.
In effect, you moved a cursor around the FAT and entered on the file you wanted to run or edit, at least on the tandy 1000. Still, I built a RPN multibase hackable calculator in 6K of code.
Where BASIC comes off the rails is that people start using it as a general programming language. Its inability to pass parameters to subroutines is easily overcome
Thus var1 = fn3130{x, v, z} can be written as:
A1=x:A2=v:A3=z:GOSUB 3130:var1=A1
In fact, once the kernel is written and documented, you can turn a generic RPN calculator script into specific special purpose code. I had mine so that all variables in the calculator start with O, P and Q. The idea was that you could write messy code outside these letters, and use the calculator as an input device.
And they say girls can't program. Ha. We just do it differently.
Regarding mnenotics, Imagine having to type something like JMP377 with the right tape loaded instead of typing in fsck.
My first computer I used was something like an 8086 (I think). The way you booted it was to load up a paper tape, and manually insert the boot sector at a specific address, and then manually press the go secuence (by loading in something like 377.
This loaded a driver for the teletype machine, which you could enter assembler codes (as numbers). Mnemotics like "a" for add and "b" for bring would have been an achievement worth speaking of.
Open source gives people the opportunity to get involved in software of their choosing. It allows for people to have a bigger say in this fast standardising world. It allows people to support "non-commercially-approved" ventures. And it allows community interests to create more adventerous progects.
I'm waiting for the day when we start to have tools that allow UI interfaces to be designed on the fly, kind of like a TeX for the UI.
People come to open source for a whole range of different reasons.
Some people want to make public some idea, but:
There is not a viable commercial market for the idea.
The person can't find someone to look after the comercial market
altristic and ethical reasons: one or more of
"ideas should be free"
"commercialisation is evil"
"should not charge for god-given ideas"
"giving" culture
tactical reasons
By giving the stuff away, you deny others the right to sell [eg MS giving away IE to kill Netscape: but they have the resources to not have to open source it.]
fear of the legal quagmire
There are lots of different reasons, just as there are lots of reasons to be at different places. Open source is an outcome, not an input. That is, it is something that one does, not a reason
NT derives from an agreement where MS would write an OS/2 where IBM would get the OEM version and MS would have some other market.
The thing was originally written as an x86 system, but because they were working on a microkernal, the thing was reasonably portable. None the less, there are 4NT versions for the alpha as well, so I think that the thing is not fully compatible.
Eventually, NT was migrated over to its new form, but the OS/2 system was of some use in the translation.
OS/2 has been prepared for the PPC as well.
Win32 is a subsystem, running on the real NT which is more akin to OS/2.
Your comment is quite amusing, and rather humourous.
I am aware of the role of a MKMSGF file. It provides teletype help to programs that need it. When you consider the amount of documentation on the OS/2 subsystem [lots] vs the number of utilities that MS currently support [none]. and then compare it against the DOS and NT programs [NT4 ships with 2 dos editors, NT2K has two also.] then the nature of the OS/2 subsystem becomes rather intriguing. What's the point in putting a subsystem in to run old OS/2 programs into a supposed rewrite.
Of course the OS/2 subsystem makes perfect sense if the system was based on OS/2, and you needed a back door to get into the kernel.
For one thing, the OS/2 system hacks directly to the kernel, and the lack of user tools for OS/2 does not imply a lack of system tools.
This in itself would explain the necessary caution given by MS in its Resource kit about disabling the OS/2 subsystem to make the system more secure.
The DOS and Win16 systems are there because Win95 can run 3.1 stuff, and NT needs that too. You can actually run drwatson under Win31. None the same, the DOS session lags keystrokes, and does not support DOSKEY.
The role of the POSIX system, which is fundementally useless as shipped, is simply that MS may advertise Posix complience. It's no more true than argueing that DOS is now part of "Windows", and OS/2 can now run "Windows" programs.
And unlike OS/2 3.x and 4.x, Windows NT4 is not going to get officially sanctioned USB drivers. Oh well.
Ah yes, "stability". I laugh when I see that. Like some of the sessions that run in it look so shaky that you wonder if the conmputer is about to reboot: programs like DOS stuff.
No, running DOS programs does not make a system unstable or insecure. But MS has imposed a penalty on NTVDM.EXE that conveys to the user that the underlying system is incredibly shaky.
As for "user mode programs bringing down the OS", hmmm - yes, I've seen charmap.exe trash the machine so hard they had to rebuild it.
The sad thing is that you HAVE to allow users to write to the OS home directory, because it stores all sorts of data goodies there. And with that, it is possible for users and applications to trash their WINNT directory. Trashed so bad that you have to rebuild the system. Oh well.
Like MS has not been weaning developers off stuffing exes and data into the OS directory, so that it could eventually be run read-only. Oh no.
I'm paraphrasing what the Resource Kit says, not what has passed C2 classifications. I have seen elsewhere that Windows is C2 complient when turned into a standalone box with no external access.
Personally, I would rather see installation routines individually C2 classified: ie if you select option 6, you get a C2 system. That means that one can reproducably install a C2 system from a given distro.
Of cause it is, they run Pacemakers on the OS/2 kernel. That's why MS were keen on using it in their new operating system.
Seriously, something that boots from the OS/2 boot sector is surely some form of hacked OS/2. They may have acquired a guy or so who wrote VMS, but the source they worked on is OS/2 1.3 through and through. Not that I am complaining.
You know, for all this alleged VMS stuff, MS is awfully quiet on it in Technet, or in the NT Help files. On the other hand, OS/2 is quite well represented in both of them.
And OS/2 support in NT is a good thing, since their networking runs on OS/2, and looks like something out of the eighties.
And, no, I'm not a MS basher. I just have lived through the reality of NT and OS/2. I am awake to it. My information does not come from MS alone. I do not believe MEGO diagrams.
eComStation is a kind of OS/2 with a Linux Distro feel to it, aka Linux and the kitchen sink.
It has lots of interesting things in it, handled through an integrated but separate installer. I like that. The installation stuff is not kept in memory every time the system boots, as is it in the Registry.
It looks cool, even. Boots off a cdrom to a GUI. Like that.
What I find distressing is that while the distro has a lot in it, it sends out a disinsentive to ISVs to compete with it. I suspect that the inclusion of IBM Works and Win-OS/2 gave OS/2 users access to word processors that ended up driving the market away from the OS/2 word processors like Describe.
What is really needed, in both the OS/2 and Windows worlds, is competing Distros. Wouldn't that be just grand.:)
Yes, Windows NT has compatibility layers. But then so does Windows 2.x, 3.x, 9x, and OS/2, Linux and so on. But the support of DOS in Windows, and the support of OS/2 in NT is fundementally different to the support of DOS in OS/2 or Linux, or POSIX or DOS in NT. In the first group, the calls are handled by the core operating system itself. In the second group, the calls are handled by a subsystem loaded when there is a need for it, ie an external plugin.
The support for OS/2 in NT is fundementally different to the DOS/Win16 or POSIX layers, even by what MS says.
The Resource Kit says that the only subsystem you have to disable to get C2 complience is the OS/2 system: ie it's the only one that calls directly to the kernel. The DOS/Win31 and POSIX systems do not call to the kernel.
NT4 kernel has ONLY OS/2 compatibility. The bulk of the operating system is not available until the shell loads and the user logs on.
All os2ss.exe and os2.exe do is shunt the calls direct to the kernel for execution. The file in question is there for OS/2 apps to call if they want console messages, it is a MKMSGF file, after all.
But the technet seems to go a fair bit into OS/2 and the subsystem, while the POSIX system is just there as an afterthought. "We support POSIX".
No, NT has correct version numbers because they follow from OS/2.
Microsoft just did not borrow "Ideas" from it. They borrowed the actual code. Look at the file c:/winnt/system32/os2/oso001.009, which contains the OS error messages. I mean, they talk about formatting disks and OS/2 boot disks there.
This system is not present in Win9x or WinME. So I would still stick with the notion that NT is a modified OS/2, probably with VMS hacks in it. But I can't see them not recycling something that works and that they can use.
Ahem NT is a rehash of OS/2. I mean, the reason they support OS/2 in NT in the first place is that the kernal runs it directly. NT3.1 actually uses an OS/2 boot sector, and NTFS is a hacked version of HPFS with the same drive type.
The reason they they need to go for a 32-bit system is that the hacks built into DOS was just not enough for their (then) expanding programs. Put simply, they exhausted DOS and were looking for help to get 32-bits under way. Hence the MS-IBM cooperation on OS/2, and NT from MS's fragment of it. FAT32 is another extention of the fs to allow the use of fat ideas into larger disks.
The original name of NT was "OS/2 NT".
It's just an evolution of an IBM product that they had code sharing rights for. Ironically, the first version of NT [ie 3.1] is correct: version 1 and 2 were the common OS/2 base.
Apparently the one true MS invention was MS BOB. It has a massive entry in technet. Enough said.
{grin} Would it be neat to have this massive solitaire engine. Solitaire is a well played game.
I understand that it needs a fifty gig partition, but System Commander can start it, so I can keep my real operating system as well.
I mean, Microsoft has come a long way in increasing the bulk of the solitaire engine from the Windows 3.0 days. We are impressed.{/grin}
And besides, to one that plays with operating systems, new operating systems really are entertainment, and hence games. Now, I just have to figure out where Windows keeps its scores.
I thought DR-DOS already had a gui: ViewMax or something. I know I ran it, but I did not find it very useful.
Back in the days when DR-DOS was stomping around, it was fairly customary to provide a character-mode menu. The idea was to get rid of as much OS as possible, to give games the space to run. The funny thing was that the DOS menu that came with my computer was more capable than the Windows Program Manager on the then just released Win31.
People who wanted more capable menus usually went out and bought third-party stuff, which was more capable than either DOSSHELL or VIEWMAX.
But I found that the DR-DOS gui mode stuff nastier than character mode stuff, especially when it's run in a window.
File Management in DOS on the other hand spawned a lot of third party stuff that is still copied to this day: eg XTree, Norton Commander.
This is slightly off topic, [in that I deal with Open Source Science, rather than the opening of source of a scientific project] but relevant.
Science and Open Source are different animals.
Just because Scientists appear to be independent agents, it does not mean that Science in itself is not cohesive.
Open Source was to some extent initiated by published listings, including source code for the early Unixen. When this was closed, the attempt was to develop "free" copies of the utilities, and then, the ultimate operating system.
The Open Source is now just moving into more advanced applications, replicating the larger commercial things in scope.
In some extent, Web Browsers were the first really open-source application that was a class leader. Netscape made a browser that rolled in the email client and ftp client, but Mosaic was there first.
But what we need is some sort of open-source idea development.
Science, especially as practiced in the West, is more based on a "religion" that has won its place through arguement, rather than cooperation.
In essence, an idea that moves into a spot becomes entrenched, and more difficult to move. So a new idea needs to arrive as more powerful to displace it through arguement.
For this reason, Scientists tend to be more cautions of letting ideas gain a foothold.
In the book "The Cathedral and the Bazaar", Eric S Raymond contrasts the world of acedemia and open source, the former competative and the latter cooperative.
Acedemia places fully developed ideas in competition, and your status in acedemia depends on your publications. The Open Source community is more open to the development of someone else's idea, and overlooks the flaws in the supplied implementation.
So whereas Scientists rejected, say, Velikovsky's ideas as crazy [because the presented package as a whole does not work], the open source people accepted Linus' alpha kernel as an idea to be extended into Linux.
We could have an open source Science that looks into alternate views. An Enthusist's science, not guided by acedemia, so to speak. But we disparage the opportunities as cranks, and build no alternate to acedemia.
de Bono offered Six Hats of thinking, as an alternate. Maybe we should look at this.
Slashdot Mirror
But if all you are after is REXX, it's better to check out Quercus REXX as their rexx multi-tasks better. It also is available as a Windows version, where you can experiment with the bugs of the Windows command line: eg passing parameters to a script, eg "REXX frog a b" does not work.
OS/2 is strong where Windows is weak, and weak where Windows is strong. It's an interesting contrast. The OS/2 community, like the Linux community, attract people who want to be there, and the latest version out this year is not dissimilar to a Linux distro for what you get (except the price tag is higher). The next version is supporting a VMWare-like project that runs Windows and Linux in a virtual computer.
The trouble that I find with MS scripting languages is that they're unusually hard and restrictive. VBA for example, makes Excel calls by a rather lengthy call, and it's not much use for the causal programer (compared to say, 1-2-3's menu-keystroke command. [Yes, I used both].
So while I try to avoid MS stuff, just seeing how the other guys do it is an eyeopener. You would know yourself, that from your unix/linux experience, you make Windows work harder, and from your Windows experience, you make Linux work harder.
For example, there are seven or eight differnt start-up objects in Windows 9x:
- msdos.sys [hidden file]
- config.sys
- autoexec.bat
- registry [many different keys]
- system.ini
- %windir%\system\vmm\*.* [just sucked up whole]
- startup folders [yes, you can have startup folders nested.
What a program is to do with a file is done in three different ways as well.It's little wonder that the thing is open to attack. You can't hunt it down unless you pretty much hack it, and follow their goofy retro thing with the 64-bit sequence: {01.22.23....}
Lack of forethought, I imagine.
EXTPROC works like this. Suppose you have a batch ADD.CMD that goes like this:
EXTPROC RPN
b %1
b %2
a
s
The command ADD 1 2 would execute the command RPN ADD 1 2, which would read each line of ADD.CMD through RPN, and give the result of 5.
That's how I read it.
Whether you let REXX or CEnvi or Perl process the RPN script is a mute point. What I was getting at was that you could create commands for different processors, AND be heard correctly.
For recollection, Linux shells do support it. And yes, I have object and regina rexx for Win/OS2/Linux :)
In the end, with the Basic RPN calculator, I used magic numbers to control the flow. This allowed for fewer labels, and robust code. The magic cookies were used to control the fan-out, with pre-testing of values (eg divide by 0) before it was attempted.
When the subroutine was done, it passed back magic cookies to indicate error codes, and its result in PA(0). It was up to the display module to announce the error message and deal with PA(0).
The message stack was also used for other things as well: you could peek at any register of the calculator.
But whatever it was, the codes were laid out so that all functions that would crash on DIV0 were together, all functions that required the range -1 to +1 were together, and so forth.
For the record, I used six magic cookies, in two sets of three. In half the cycle, set A would be controlling the flow, and set B were scratch data, while the other half B controls the flow, and A is the scratch data.
This greatly reduced the number of GOTO and GOSUB lines. I think in the end, there were 10 different labels of major jumps.
As for tweaking code: I still do it to great effect. I can still pull 8x increase in speed by selecting the correct algorithms.
Now, with Beowulf clusters, Distributed computing, and IBM runing Linux in mainframe VM machines, one is seeing lots of little PCs competing on mainframe turf. PCs have largely displaced mainframe terminals. A pc running a 3270 emulator is cheaper than a 3270 terminal!
Still, of the four you quote, one bemoans the change of data and information. Another just agrees with what someone else's correction, a third is a hi to some troll, and the fourth is the story. This makes five.
I can find a lot to do in the evenings, thank you. It's just morning over here: nearly lunch time, actually. And yes, I am currently bored, apart from passing on my experiences in the old times.
I'd be more likely to while the hours away on Civ3 or REXX than read Ann Rice novels.
In OS/2 it is easy to do that. This is because of EXTPROC and REXX. From that, you can write a REXX script for the processor, and feed the language script in as data. Saved many hundreds of hours' work doing it like that.
Maybe you can do that in Linux too - don't know enough about that.
Most good authorities on computer languages says you should learn several languages. This allows you to think more creatively. I know.
It was years ago. Haven't seen the docos for 15 years. You may be right...
I wonder how many IT people suggest technologies that are not computer-related: eg how many people suggest paper cards as a solution. I know I have.
You see, once you start fiddling around with the hardware like Betty H did, you start using it wisely. It is one of the reasons that Unix works so well.
My computer at college in 1981 was something nearing the end of its life. It was an 8086 with 4K of ram, and a paper tape drive. To boot it, you load up the tape, and load three values into ram (by a series of eight switches and a "set" switch), and then send a command 377 to the processor. This would jump it tot a location in memory, and then run the commands that you loaded there (effectively JMP address), which would then run the KEX program. KEX was a driver for a teletype. After that, you input through the keyboard by assembler code.
Compared to that, mnenomics like a for add and b for bring would have been a godsend.
Of fortran, basic and cobol. In the days of wire wound core, each bit of the byte made the machine more expensive, and there was some comprimise on the size of the bit. Fortran was designed to run on a six-bit machine. Even Knuth's MIX is underpowered by modern computers.
BASIC is intended to run in small memory. MS made their packet by bumming it into 4K of ram, with a point and shoot interface.
In effect, you moved a cursor around the FAT and entered on the file you wanted to run or edit, at least on the tandy 1000. Still, I built a RPN multibase hackable calculator in 6K of code.
Where BASIC comes off the rails is that people start using it as a general programming language. Its inability to pass parameters to subroutines is easily overcome
Thus var1 = fn3130{x, v, z} can be written as:
A1=x:A2=v:A3=z:GOSUB 3130:var1=A1
In fact, once the kernel is written and documented, you can turn a generic RPN calculator script into specific special purpose code. I had mine so that all variables in the calculator start with O, P and Q. The idea was that you could write messy code outside these letters, and use the calculator as an input device.
And they say girls can't program. Ha. We just do it differently.
My first computer I used was something like an 8086 (I think). The way you booted it was to load up a paper tape, and manually insert the boot sector at a specific address, and then manually press the go secuence (by loading in something like 377.
This loaded a driver for the teletype machine, which you could enter assembler codes (as numbers). Mnemotics like "a" for add and "b" for bring would have been an achievement worth speaking of.
I'm waiting for the day when we start to have tools that allow UI interfaces to be designed on the fly, kind of like a TeX for the UI.
- Some people want to make public some idea, but:
- There is not a viable commercial market for the idea.
- The person can't find someone to look after the comercial market
- altristic and ethical reasons: one or more of
- "ideas should be free"
- "commercialisation is evil"
- "should not charge for god-given ideas"
- "giving" culture
- tactical reasons
- By giving the stuff away, you deny others the right to sell [eg MS giving away IE to kill Netscape: but they have the resources to not have to open source it.]
- fear of the legal quagmire
There are lots of different reasons, just as there are lots of reasons to be at different places. Open source is an outcome, not an input. That is, it is something that one does, not a reasonThe thing was originally written as an x86 system, but because they were working on a microkernal, the thing was reasonably portable. None the less, there are 4NT versions for the alpha as well, so I think that the thing is not fully compatible.
Eventually, NT was migrated over to its new form, but the OS/2 system was of some use in the translation.
OS/2 has been prepared for the PPC as well.
Win32 is a subsystem, running on the real NT which is more akin to OS/2.
I am aware of the role of a MKMSGF file. It provides teletype help to programs that need it. When you consider the amount of documentation on the OS/2 subsystem [lots] vs the number of utilities that MS currently support [none]. and then compare it against the DOS and NT programs [NT4 ships with 2 dos editors, NT2K has two also.] then the nature of the OS/2 subsystem becomes rather intriguing. What's the point in putting a subsystem in to run old OS/2 programs into a supposed rewrite.
Of course the OS/2 subsystem makes perfect sense if the system was based on OS/2, and you needed a back door to get into the kernel. For one thing, the OS/2 system hacks directly to the kernel, and the lack of user tools for OS/2 does not imply a lack of system tools.
This in itself would explain the necessary caution given by MS in its Resource kit about disabling the OS/2 subsystem to make the system more secure. The DOS and Win16 systems are there because Win95 can run 3.1 stuff, and NT needs that too. You can actually run drwatson under Win31. None the same, the DOS session lags keystrokes, and does not support DOSKEY.
The role of the POSIX system, which is fundementally useless as shipped, is simply that MS may advertise Posix complience. It's no more true than argueing that DOS is now part of "Windows", and OS/2 can now run "Windows" programs.
Ah yes, "stability". I laugh when I see that. Like some of the sessions that run in it look so shaky that you wonder if the conmputer is about to reboot: programs like DOS stuff.
No, running DOS programs does not make a system unstable or insecure. But MS has imposed a penalty on NTVDM.EXE that conveys to the user that the underlying system is incredibly shaky.
As for "user mode programs bringing down the OS", hmmm - yes, I've seen charmap.exe trash the machine so hard they had to rebuild it.
The sad thing is that you HAVE to allow users to write to the OS home directory, because it stores all sorts of data goodies there. And with that, it is possible for users and applications to trash their WINNT directory. Trashed so bad that you have to rebuild the system. Oh well.
Like MS has not been weaning developers off stuffing exes and data into the OS directory, so that it could eventually be run read-only. Oh no.
Personally, I would rather see installation routines individually C2 classified: ie if you select option 6, you get a C2 system. That means that one can reproducably install a C2 system from a given distro.
Seriously, something that boots from the OS/2 boot sector is surely some form of hacked OS/2. They may have acquired a guy or so who wrote VMS, but the source they worked on is OS/2 1.3 through and through. Not that I am complaining.
You know, for all this alleged VMS stuff, MS is awfully quiet on it in Technet, or in the NT Help files. On the other hand, OS/2 is quite well represented in both of them.
And OS/2 support in NT is a good thing, since their networking runs on OS/2, and looks like something out of the eighties.
And, no, I'm not a MS basher. I just have lived through the reality of NT and OS/2. I am awake to it. My information does not come from MS alone. I do not believe MEGO diagrams.
It has lots of interesting things in it, handled through an integrated but separate installer. I like that. The installation stuff is not kept in memory every time the system boots, as is it in the Registry.
It looks cool, even. Boots off a cdrom to a GUI. Like that.
What I find distressing is that while the distro has a lot in it, it sends out a disinsentive to ISVs to compete with it. I suspect that the inclusion of IBM Works and Win-OS/2 gave OS/2 users access to word processors that ended up driving the market away from the OS/2 word processors like Describe.
What is really needed, in both the OS/2 and Windows worlds, is competing Distros. Wouldn't that be just grand. :)
The support for OS/2 in NT is fundementally different to the DOS/Win16 or POSIX layers, even by what MS says.
The Resource Kit says that the only subsystem you have to disable to get C2 complience is the OS/2 system: ie it's the only one that calls directly to the kernel. The DOS/Win31 and POSIX systems do not call to the kernel.
NT4 kernel has ONLY OS/2 compatibility. The bulk of the operating system is not available until the shell loads and the user logs on.
All os2ss.exe and os2.exe do is shunt the calls direct to the kernel for execution. The file in question is there for OS/2 apps to call if they want console messages, it is a MKMSGF file, after all.
But the technet seems to go a fair bit into OS/2 and the subsystem, while the POSIX system is just there as an afterthought. "We support POSIX".
No, NT has correct version numbers because they follow from OS/2.
This system is not present in Win9x or WinME. So I would still stick with the notion that NT is a modified OS/2, probably with VMS hacks in it. But I can't see them not recycling something that works and that they can use.
The reason they they need to go for a 32-bit system is that the hacks built into DOS was just not enough for their (then) expanding programs. Put simply, they exhausted DOS and were looking for help to get 32-bits under way. Hence the MS-IBM cooperation on OS/2, and NT from MS's fragment of it. FAT32 is another extention of the fs to allow the use of fat ideas into larger disks.
The original name of NT was "OS/2 NT". It's just an evolution of an IBM product that they had code sharing rights for. Ironically, the first version of NT [ie 3.1] is correct: version 1 and 2 were the common OS/2 base.
Apparently the one true MS invention was MS BOB. It has a massive entry in technet. Enough said.
I understand that it needs a fifty gig partition, but System Commander can start it, so I can keep my real operating system as well.
I mean, Microsoft has come a long way in increasing the bulk of the solitaire engine from the Windows 3.0 days. We are impressed.{/grin}
And besides, to one that plays with operating systems, new operating systems really are entertainment, and hence games. Now, I just have to figure out where Windows keeps its scores.
Back in the days when DR-DOS was stomping around, it was fairly customary to provide a character-mode menu. The idea was to get rid of as much OS as possible, to give games the space to run. The funny thing was that the DOS menu that came with my computer was more capable than the Windows Program Manager on the then just released Win31.
People who wanted more capable menus usually went out and bought third-party stuff, which was more capable than either DOSSHELL or VIEWMAX.
But I found that the DR-DOS gui mode stuff nastier than character mode stuff, especially when it's run in a window.
File Management in DOS on the other hand spawned a lot of third party stuff that is still copied to this day: eg XTree, Norton Commander.
Science and Open Source are different animals.
Just because Scientists appear to be independent agents, it does not mean that Science in itself is not cohesive.
Open Source was to some extent initiated by published listings, including source code for the early Unixen. When this was closed, the attempt was to develop "free" copies of the utilities, and then, the ultimate operating system.
The Open Source is now just moving into more advanced applications, replicating the larger commercial things in scope.
In some extent, Web Browsers were the first really open-source application that was a class leader. Netscape made a browser that rolled in the email client and ftp client, but Mosaic was there first.
But what we need is some sort of open-source idea development.
Science, especially as practiced in the West, is more based on a "religion" that has won its place through arguement, rather than cooperation.
In essence, an idea that moves into a spot becomes entrenched, and more difficult to move. So a new idea needs to arrive as more powerful to displace it through arguement.
For this reason, Scientists tend to be more cautions of letting ideas gain a foothold.
In the book "The Cathedral and the Bazaar", Eric S Raymond contrasts the world of acedemia and open source, the former competative and the latter cooperative.
Acedemia places fully developed ideas in competition, and your status in acedemia depends on your publications. The Open Source community is more open to the development of someone else's idea, and overlooks the flaws in the supplied implementation.
So whereas Scientists rejected, say, Velikovsky's ideas as crazy [because the presented package as a whole does not work], the open source people accepted Linus' alpha kernel as an idea to be extended into Linux.
We could have an open source Science that looks into alternate views. An Enthusist's science, not guided by acedemia, so to speak. But we disparage the opportunities as cranks, and build no alternate to acedemia.
de Bono offered Six Hats of thinking, as an alternate. Maybe we should look at this.