DOS ain't dead

September 10, 2007:

> The Free Pascal Compiler team is pleased to announce the
> release of FPC 2.2.0!

Website: http://www.freepascal.org/
Download: http://www.freepascal.org/download.var

Changes (from WHATSNEW.TXT):
> GO32V2 DOS platform got a long needed update

---
Know your limits.h

But not for DOS.

> But not for DOS.

Because of a lack of release builders and testers, 2.2.0 is
only available for a limited number of platforms and not
in all package formats. If you want to change this and
build and test future releases, contact us via the mailing lists.

I can't test everything!! IMO, they need to ask in the appropriate places (e.g. comp.os.msdos.djgpp) or even at the FreeBASIC forum.

---
Know your limits.h

> > But not for DOS.
>
> Because of a lack of release builders and testers, 2.2.0 is
> only available for a limited number of platforms and not
> in all package formats. If you want to change this and
> build and test future releases, contact us via the mailing lists.

And yet, if you look at the FTP site ... :

* fpc-2.0.5dos.zip     ZIP Archive     10,036 KB       09/11/2007 02:23:00 AM
* fpc-2.2.1.go32v2.zip ZIP Archive     11,561 KB       09/10/2007 02:02:00 AM
* fpc-2.3.1.go32v2.zip ZIP Archive     12,239 KB       09/10/2007 02:44:00 AM

---
Know your limits.h

> > Free Pascal Compiler team is pleased to announce the release of FPC 2.2.0
> > GO32V2 DOS platform got a long needed update

Did or didn't ? Is the memory alloc bug fixed ? Awaiting test report from Laaca

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> And yet, if you look at the FTP site ... :

OK, snapshots of binaries-only packages. But wouldn't a fully tested, complete upgrade package be nicer?

> wouldn't a fully tested, complete upgrade package be nicer?

YES. But they just don't have DOS

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> I can't test everything!! IMO, they need to ask in the
> appropriate places (e.g.

Maybe we should bundle our forces?

EDIT: Their build system is crap squared. I even tried to cross-compile using the Win32 release w/o luck.

EDIT (2): Got it halfway working.

> comp.os.msdos.djgpp)
> or even at the FreeBASIC forum.

Why should a BASIC or C/C++ guy touch "boring" Pascal?

---
Forum admin

> Why should a BASIC or C/C++ guy touch "boring" Pascal?

The DOS ports of these three compilers are all based upon the DJGPP toolchain (AS, GDB, LD, CWSDPMI, etc) and libraries (LIBC, LIBEMU) even if the "compiler proper" of each (FPC.EXE, FBC.EXE) are currently compiled/written in themselves.

So, anybody working with one should already have some familiarity with the others.

---
Know your limits.h

> DOS ports of these three compilers are all based upon the DJGPP toolchain AS, GDB, LD

YES.

> CWSDPMI

Very true and very obsolete.

> and libraries (LIBC, LIBEMU) even

Doesn't FPC have it's own (buggy) libraries and (buggy) startup code ?

> they need to ask in the appropriate places (comp.os.msdog.djgpp) or even at the FreeBASIC

What's needed is a place with people interested in PASCAL and having DOS ... so far I know exactly 1 (one) person qualifying (definitely about 1st, hopefully (still ?) on 2nd )

On comp.os.msdog.djgpp, people don't have DOS

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> > CWSDPMI
>
> Very true and very obsolete.

It's not obsolete, because it works very well in most situations.

> > and libraries (LIBC, LIBEMU) even
>
> Doesn't FPC have it's own (buggy) libraries and (buggy) startup code ?

Why do you always guess or repeat other's opinions? Investigate and draw your own conclusions.

> What's needed is a place with people interested in PASCAL and

It's spelled Pascal after Blaise Pascal.

> having DOS ... so far I know exactly 1 (one) person
> qualifying (definitely about 1st, hopefully (still ?) on 2nd )

Laaca?

> On comp.os.msdog.djgpp, people don't have DOS

That's wrong.

---
Forum admin

> But not for DOS.

Please try http://www.bttr-software.de/misc/fpc/!
It still lacks debugger support in text-mode IDE, but I will fix this "in the next days".

---
Forum admin

> > CWSDPMI
>
> Very true and very obsolete.

Not obsolete because it still works, but there are other options. However, it does swapping (whether you like it or not), which can indeed be useful on much older machines with really low memory.

> Doesn't FPC have it's own (buggy) libraries and (buggy) startup code ?

I was mainly referring to FBC (which has such libraries), didn't look at FPC too closely. But I know FPC uses some DJGPP stuff, so that was the point: they're all related.

> What's needed is a place with people interested in PASCAL and
> having DOS ... so far I know exactly 1 (one) person
> qualifying (definitely about 1st, hopefully (still ?) on 2nd )

The real question is "Why is everything all about Windows and Linux?" What is accomplished by writing anything for them that can't be done in DOS as well?

> On comp.os.msdog.djgpp, people don't have DOS

I dunno, some use Win98SE, some use WinXP, it's all related to how old your computers are, what OSes came preinstalled (the main way people get Windows), etc.

At least with a FreeDOS bootdisk, people can easily use DOS without any hassles (even under QEMU if no floppy drive is available).

---
Know your limits.h

> Please try http://www.bttr-software.de/misc/fpc/!

I tried dos220_pre2. First problem: Accepts being installed on drive C:\ only. Couldn't find other drives.

Most EXEs (all but CWSDPMI) are compressed with UPX, some with v1,25, some with v1.91 (including the suppplied v1.25 itself). UPX v1.25 does not decompress files that are compressed with v1.91.

Includes Info-ZIP v2.2 - going to current v2.32 would be a big improvement (built-in encryption, etc.)

IDE: No Help.

> I tried dos220_pre2. First problem: Accepts being installed on drive C:\
> only. Couldn't find other drives.

Do you get "The drive D: is not valid."? Press "n" twice to continue. If it doesn't help, you have to tell more about your OS and your partitions.

> Most EXEs (all but CWSDPMI) are compressed with UPX, some with v1,25, some
> with v1.91 (including the suppplied v1.25 itself). UPX v1.25 does not
> decompress files that are compressed with v1.91.

But this is not especially related to release 2.2.0. I agree to the FPC core team to change that for later versions, but not for 2.2.0.

> Includes Info-ZIP v2.2 - going to current v2.32 would be a big improvement
> (built-in encryption, etc.)

see above

> IDE: No Help.

That requires `doc-html.zip', which is part of still pending `dos220_full.zip'.

---
Forum admin

> > I tried dos220_pre2. First problem: Accepts being installed on drive C:\
> > only. Couldn't find other drives.
>
> Do you get "The drive D: is not valid."? Press "n" twice to continue. If
> it doesn't help, you have to tell more about your OS and your partitions.

OK, that worked - in effect, it can be told to install on an invalid drive, and it will. BTW, the problem occurred with both NTFS and FAT drives.

> > Most EXEs (all but CWSDPMI) are compressed with UPX, some with v1,25,
> > some with v1.91 (including the suppplied v1.25 itself).
>
> But this is not especially related to release 2.2.0. I agree to the FPC
> core team to change that for later versions, but not for 2.2.0.
>
> > Includes Info-ZIP v2.2
>
> see above

Understood. A strange situation - is compressing UPX with a later version of itself a new form of encryption?

> > IDE: No Help.
>
> That requires `doc-html.zip', which is part of still pending
> `dos220_full.zip'.

OK, I downloaded doc-html.tar.gz, with same contents, both separately available. The HTML files install in the IDE, but look like crap - viewer is a narrow box, cuts off line ends.

I've never used FreePas before, and must say I'm not encouraged by the IDE - it needs a lot of work. Considering how old the project is, it's ridiculous. What else is lurking under the surface?

I downloaded the testing prerelease version of FPC 2.2.0 for DOS by Robert Riebisch and tried to compile Blocek.

From my mail to Robert:

"Problem is when compiling Blocek in Release mode.
(I set the compilation directives in this way: If you compile in Debug or
Normal mode, Blocek compiles without PASJPEG so it has no JPG image support.
If you use Release mode, JPG support is enabled.)
So, problem occurs while compilation on line 325 of jmemmgr.pas file in
.\PASJPEG directory. I have no idea what to do with it. Just try it yourself

In FPC 2.1.4 (2.2.0 - release candidate1) is everything OK.

Note: I use slightly modified PASJPEG library from the standard one. The
difference is that I use streams for file I/O and the standard one uses
normal pascal routines."

Conclusion: 2.2.0 it is a regression from 2.1.4

> > Very true and very obsolete.
> It's not obsolete, because it works very well in most situations.

It performs best when you install the compiler in NTVDM or similar "DOS" where you "already have a DPMI server" and CWSDPMI gets fully ignored

> Why do you always guess or repeat other's opinions?
> > What's needed is a place with people interested in PASCAL
> It's spelled Pascal

Exactly as I wrote.

> Laaca?

Good guess

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> Conclusion: 2.2.0 it is a regression from 2.1.4

Thanks. The truth is out

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

I did some more investigation of the new bug in PASJPEG compilation, sent it to FPC team and they took it very serious and want to fix it soon:
http://www.freepascal.org/mantis/view.php?id=9706

Anyway, I built Blocek with FPC 2.2.0 (with JPG support). Can you tell me whether stability increased or not?
http://laaca-mirror.ic.cz/blo220.rar

> I've never used FreePas before, and must say I'm not encouraged by the IDE
> - it needs a lot of work. Considering how old the project is, it's
> ridiculous. What else is lurking under the surface?

Obviously, they support a LOT of platforms (including Win64), so DOS is not a huge priority to them, for whatever reason (lack of time or developers, probably, but maybe not). Whatever, they are still to be applauded for their accomplishments. Very impressive, IMO. (I like compilers, heh.)

Sometimes I wonder if people avoid DOS because either they want everyone to (have to) use GNU/Linux or they think DOS always means MS-DOS (and hate it because of them). I mean, FreeDOS is more "free" than any Windows, and yet people always port stuff to Windows and not nearly as much to DOS (though we ain't truly forgotten, just overlooked).

---
Know your limits.h

> Obviously, they support a LOT of platforms (including Win64), so DOS is
> not a huge priority to them, for whatever reason (lack of time or
> developers, probably, but maybe not). Whatever, they are still to be

BOTH! That's why I try to help them.

> Sometimes I wonder if people avoid DOS because either they want everyone
> to (have to) use GNU/Linux or they think DOS always means MS-DOS (and hate
> it because of them). I mean, FreeDOS is more "free" than any Windows, and

Why would anyone hate MS-DOS? It always worked very well for me.

> yet people always port stuff to Windows and not nearly as much to DOS
> (though we ain't truly forgotten, just overlooked).

Therefore we should strike back with FreeDOS 1.1.

---
Forum admin

> Why would anyone hate MS-DOS? It always worked very well for me.

MS = Microsoft. Everywhere you turn these days, MS pokes its head. They write (hire?) a LOT of software. And they get a lot of backhanded "publicity" from Linux users (MS this, MS that, blah blah, "M$ sux0rz", etc).

> Therefore we should strike back with FreeDOS 1.1.

In progress! (albeit slowly)

P.S. regarding this:

> If you try to compile PASJPEG with no optimalizations, all is
> OK. If you turn on Level 1 optimalizatios, compilation ends
> on JMEMMGR.PAS on line 325 with error report: "asm: [push imm8]
> invalid combination of opcode and operands"

I found no asm blocks in that file, so I dunno, but obviously you can't push a byte (only a word or larger). "PUSH 8" would be either "6A 08" (push 0008) or "66 6A 08" (push 00000008) in e.g. 16-bit mode.

---
Know your limits.h

> > Why would anyone hate MS-DOS? It always worked very well for me.
>
> MS = Microsoft. Everywhere you turn these days, MS pokes its head. They
> write (hire?) a LOT of software. And they get a lot of backhanded

You didn't get the point: It doesn't explain, why anyone would port software to (Microsoft) Windows, if he already hates (Microsoft) DOS.

> "publicity" from Linux users (MS this, MS that, blah blah, "M$ sux0rz",
> etc).

Or from guys like DOS386.

---
Forum admin

> You didn't get the point: It doesn't explain, why anyone would port
> software to (Microsoft) Windows, if he already hates (Microsoft) DOS.

Windows does do some things DOS can't (else nobody would use it, duh), mostly due to more commercial support, more developers, tons of device drivers, etc. As far as using it instead of DOS? Money is the only reason I can think of (maybe fame, too?). There's a large userbase of Windows out there (vs. 3% or so for Mac OS X or whatever for GNU/Linux). Since DOS is commercially dead (more or less), some people refuse to use it. (Or course, they find all kinds of nonsensical reasons to explain their decisions, usually the same old crud that's been a non-issue for years now.)

> > Obviously, they support a LOT of platforms (including Win64), so DOS is
> > not a huge priority to them, for whatever reason (lack of

... interest. The reason are all the other "much better" platforms

> > Sometimes I wonder if people avoid DOS because either they want everyone
> > to (have to) use GNU/Linux

YES.

> > or they think DOS always means MS-DOS (and hate it because of them).

YES.

> I mean, FreeDOS is more "free" than any Windows, and

... nevertheless you tested the Galactical stuff on Vi$ta

> Why would anyone hate MS-DOS? It always worked very well for me.

I do hate MS-DOG Why ? What is MS-DOG (or MS-DOS, M$-DO$, messy-dos, ...) at all ? Anyone has a consistent definition ? I don't ...

> > yet people always port stuff to Windows and not nearly as much to DOS

Because they don't have DOS, and consider DOS as desperately old, obsolete, dead, crappy, unusable, no-GUI, no-multitask, no-multiabuser, no-internet, no-multimedia, no-sound-in-DOS, not-free, M$-proprietary, DOS=MS-DOG, no-hardware-support, runs-on-oldest-PC's-only, no-serious-OS, "Windows"-booter only, needs-"Windows"-anyway, 16-bit-only, 640-KiB-only, unreliable-in-NTVDM, no-GUI-in-Vi$ta, no-DOS-at-all-in-Vi$ta-64, no-USB-in-DOS, beginner-programmers-toy-only, ...

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> I did some more investigation of the new bug in PASJPEG compilation, sent
> it to FPC team and they took it very serious and want to fix it soon:
> http://www.freepascal.org/mantis/view.php?id=9706

COOL. F.K. (used to care about FPC for DOS, but it's 10 years ago ) changed the priority to "urgent"

> http://laaca-mirror.ic.cz/blo220.rar

Occasion to test BTTR's UNRAR - passed

> Can you tell me whether stability increased or not?

OK, tested:

- No regressions found
- JPG seems to work (I don't have many of them )
- PNG decoder is buggy (bug probably not new, and suspiciously similar to Arachne's PNG decoder bug - messes up low-color palletized PNG's)
- My evil 15 MiB text loads
- S&R securely crashes, even on small texts (as always before)
- Compilation date is wrong (don't you pick it from the clock ?)
- Memory leak/alloc bug not yet verified (too much RAM and too little time so far)

> but obviously you can't push a byte

IIRC 8080 was the last CPU with this feature

> "PUSH 8" would be either "6A 08" (push 0008)

Operand occupies 8 but it still pushes 32 bits

> or "66 6A 08" (push 00000008) in e.g. 16-bit mode.

Maybe ... or "68 08 00 00 00"

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> Most EXEs (all but CWSDPMI) are compressed with UPX, some with v1,25, some
> with v1.91 (including the suppplied v1.25 itself). UPX v1.25 does not
> decompress files that are compressed with v1.91.

Bad They failed to read UPX docs - UPX x.9x versions are intended for testing only, not for releases Next time they should use UPX 3.01 or even better none at all

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> Because they don't have DOS, and consider DOS as desperately old,
> obsolete, dead, crappy, unusable, no-GUI, no-multitask, no-multiabuser,
> no-internet, no-multimedia, no-sound-in-DOS, not-free, M$-proprietary,
> DOS=MS-DOG, no-hardware-support, runs-on-oldest-PC's-only, no-serious-OS,
> "Windows"-booter only, needs-"Windows"-anyway, 16-bit-only, 640-KiB-only,
> unreliable-in-NTVDM, no-GUI-in-Vi$ta, no-DOS-at-all-in-Vi$ta-64,
> no-USB-in-DOS, beginner-programmers-toy-only, ...

Why are you here?

> I do hate MS-DOG Why ? What is MS-DOG (or MS-DOS, M$-DO$, messy-dos,
> ...) at all ? Anyone has a consistent definition ? I don't ...

I see. Just "feelings", no substantiation. Why are you here? You sound like a typical Linux kiddie...

---
Forum admin

> > I mean, FreeDOS is more "free" than any Windows, and
>
> ... nevertheless you tested the Galactical stuff on Vi$ta

Only after testing in real DOS. (Hey, I'm human, I don't use DOS exclusively, just a really big fan.) Besides, somebody has to test it under Windows since it's so widespread.

> > > yet people always port stuff to Windows and not nearly as much to DOS
>
> Because they don't have DOS, and consider DOS as desperately old,
> obsolete, dead, crappy, unusable, no-GUI, no-multitask, no-multiabuser,
> no-internet, no-multimedia, no-sound-in-DOS, not-free, M$-proprietary,
> DOS=MS-DOG, no-hardware-support, runs-on-oldest-PC's-only, no-serious-OS,
> "Windows"-booter only, needs-"Windows"-anyway, 16-bit-only, 640-KiB-only,
> unreliable-in-NTVDM, no-GUI-in-Vi$ta, no-DOS-at-all-in-Vi$ta-64,
> no-USB-in-DOS, beginner-programmers-toy-only, ...

Don't have DOS? Get FreeDOS! (BASE .ISO is only 8 MB, bootdisk is 1.44 MB)

USB? We need more drivers (ported from Linux? I wish ...). GUI? I prefer TUI. Multitask? Again, probably need more developers (or testers for TriDOS, which ain't quite stable, supposedly; MT is also kinda wimpy but interesting too). No sound? Blame AC97 for breaking compatibility with old SB (use Mpxplay or Open Cubic Player or emulation drivers, if you can). Runs on older PCs? That's a pro, not a con! 16-bit? So what, if it works it works. Actually, most DOS software requires 386+, that's kinda implied these days. 640k? You can obviously easily use more than that. NTVDM/Vista/Win64 ain't real DOS anyways (nor is DOSEMU but better than nothing, I guess).

> > Because they don't have DOS, and consider DOS as desperately old,
> > obsolete, dead, crappy, unusable ...
>
> Why are you here?

He is obviously not of that opinion, he likes DOS (obviously?), just explaining some other people's weird reasoning.

> > I do hate MS-DOG
>
> I see. Just "feelings", no substantiation. Why are you here? You sound
> like a typical Linux kiddie...

You know as well as I do that he complains about certain things (and rightfully so, IMO) that could/should be easily fixed or be non-issues. He doesn't idolize Linux (though others do) as the perfect solution for all machines (ancient, "older", or even newer cpus).

Don't be so sensitive.

---
Know your limits.h

> Don't be so sensitive.

Sorry!

---
Forum admin

Hi DOS386,
can you test Blocek again?
I think I solved a crash which occured when you
1) loaded huge text which filled all RAM so swapping was in action
2) did replace/change all/without confirming
3) pressed some key

In keyboard interrupt routine I didn't lock all touched variables so it could happen that after keypress could be called interrupt routine which wasn't in memory in thist moment but on swap file on disk.
Now it should be fixed - download Blocek 1.32c

http://www.laaca-mirror.ic.cz/

> Please try http://www.bttr-software.de/misc/fpc/!
> It still lacks debugger support in text-mode IDE, but I will fix this "in
> the next days".

I uploaded a new pre-release version. Please tell me your results for debugging applications inside IDE!

Next version will probably have Pierre's (?) non-LFN fix.

---
Forum admin

Steve and rr (really needed 2x ? ) wrote:

> Why are you here?

No idea

> You sound like a typical Linux kiddie...

Please read my posts more carefully in the future. I don't have Loonix, and it didn't work in my test long ago anyway

BTW: I collected other people's claims in the "hot" post, not mine

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> can you test Blocek again?

Will ASAP - sorry lack of time and access to world's only (=my) DOS PC

> I think I solved a crash which occured when you

Well, it occurred with small texts and HDPMI32 as well

> Now it should be fixed - download Blocek 1.32c
> http://www.laaca-mirror.ic.cz/

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

> > You sound like a typical Linux kiddie...
>
> Please read my posts more carefully in the future. I don't have Loonix,
> and it didn't work in my test long ago anyway

I already knew from Rugxulo's reply. Sorry again! Welcome back!

---
Forum admin

> can you test Blocek again?
> I think I solved a crash which occured when you
> 1) loaded huge text which filled all RAM so swapping was
> 2) did replace/change all/without confirming
> Now it should be fixed - download Blocek 1.32c
> http://www.laaca-mirror.ic.cz/

It isn't Also it crashes (as it always did) with small texts already ...

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***