DOS ain't dead

Hi,

HX has also been updated, now at v2.13.

What's new:
- it runs Qemu 0.91
- more accurate timer implementation, might improve speed and sound quality of some SDL apps (incl Qemu + DosBox)
- works in VirtualBox
- some bugfixes in HDPMI

http://www.japheth.de/hx.html

---
MS-DOS forever!

> HX has been updated, v2.13

Thanks

> What's new:
> - it runs Qemu 0.91
> - more accurate timer implementation, might improve speed and sound
> - some bugfixes in HDPMI

More:

- Huge amount of "dummy" + a few working imports added

- "Enh-Meta" added, less missing imports in NCONVERT, some remaining:

dpmild32: import not found: GetCharABCWidthsFloatA
dpmild32: import not found: StrokePath
dpmild32: import not found: FillPath
dpmild32: import not found: EndPath
dpmild32: import not found: BeginPath
dpmild32: import not found: GetEnhMetaFileA
dpmild32: file GDI32.dll

- Both DPMILD32.BIN and HDLD32.BIN are now contained in HXDEV, also HDPMI32.INC is there - but seems to be the 2007-Sep version

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

> - Both DPMILD32.BIN and HDLD32.BIN are now contained in
> HXDEV, also HDPMI32.INC is there - but seems to be the 2007-Sep
> version

You can create your own version. See hdpmi32.mak in hxsrc.

Or use this: http://www.japheth.de/Download/hdpmi32.zip

---
MS-DOS forever!

Wengier Wu made me aware that the special version of hdpmi32 which runs clients with IOPL 0 - which is needed by Virtual PC - hasn't been updated yet. This has been fixed: http://www.japheth.de/Download/hdpmivpc.zip

---
MS-DOS forever!

> Or use this



> HX has been updated, now at 2.13

Not fixed:

- OPTIPNG-bug

- MPLAYER 1.0RC2 doesn't work. This is not a regression of HX, but a "regression" of MPLAYER. SVN from cca 2007 March to September did work (not excellently), 1.0RC2 does no longer. Anyone else tested ? Of course, the relevance of this is almost ZERO , we have Mik's much better port

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

> 1.0RC2 does no longer. Anyone else tested ? Of course,
> the relevance of this is almost ZERO , we have Mik's much
> better port

Tested, does not work either. However, it's still somewhat useful over Mik's port, as Mik's port does not include some codecs (at least for now), e.g. codec for RMVB videos.

the HDLD32.BIN stub binary in hxdev 2.13 is buggy.

A fix is available at

http://www.japheth.de/Download/hdld32.zip

---
MS-DOS forever!

> A fix is available at

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

> HX has also been updated, now at v2.13

> 01/14/2008: HDEdit version 1.11

COOL.

> - HDEDITW.EXE the HDEDIT binary for Windows XP

Good occasion to test DKRNL32's XP physical disk device emulation -> crash

Not that it would be extremely useful to run the XP version on DOS when we have a native one ... but why does it crash ?

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

> Not that it would be extremely useful to run the XP version on DOS when we
> have a native one ... but why does it crash ?

Because there was an alignment bug in the support of IOCTL_DISK_GET_DRIVE_GEOMETRY_EX. Qemu uses this function just to get the disk size, which worked fine, but hdeditw must use it to get the Int 13h extensions, and this won't work with dkrnl32 of hxrt v2.13. It has been fixed in the meantime and hdeditw works perfectly in DOS now (but won't in Win9x ).

---
MS-DOS forever!

to provide a tool which can debug HDLD32.BIN an update of DEBUG(x) was necessary:

http://www.japheth.de/Download/Debug110.zip

---
MS-DOS forever!

Hi,

there is a bugfix release of hxrt available. None of the fixed bugs is critical.

HDPMI:
- CR0 MP and EM bits are now cleared for the first client only. This allows to implement an FP emulator DPMI TSR.

DKRNL32:
- bugfix: IOCTL_DISK_GET_DRIVE_GEOMETRY_EX didn't return the drive parameters due to an alignment error.
- bugfix: dkrnl's exception handler destroyed content of EAX.
- bugfix: SearchPath() no longer fails if a full path is given as filename.
- bugfix: DKRNL32 initialization disabled a possibly installed FPU emulator.


http://www.japheth.de/Download/hxrt.zip

---
MS-DOS forever!

> http://www.japheth.de/Download/hxrt.zip

Doesn't work - upper case required:
http://www.japheth.de/Download/HXRT.ZIP

> Doesn't work - upper case required:
> http://www.japheth.de/Download/HXRT.ZIP

Fixed. It's not the full filename, just the extension. If I pack the stuff in Windows or in DOS with DOSLFN loaded, I get a ".zip", but in pure DOS a ".ZIP" extension. And the Unix host is too stupid to realize what I really want to do when uploading.

---
MS-DOS forever!

> And the Unix host is too stupid to realize what I really want to do when
> uploading.

Thank God it doesn't try to be "clever" like Windows systems.

---
Forum admin

> Fixed. It's not the full filename, just the extension.

The HX DOS page displays the whole filename in upper case, and delivers it the same way. I have now tried feeding the link to my browser with hxrt.ZIP, and it worked - still getting HXRT.ZIP

> If I pack the stuff
> in Windows or in DOS with DOSLFN loaded, I get a ".zip", but in pure DOS a
> ".ZIP" extension.

I have noticed mixed case filenames for other packages - now I know their history.

> And the Unix host is too stupid to realize what I really
> want to do when uploading.

Right.

Don't you just case-sensitive file systems, not!

Now FD Debug has arrived at v1.11:

http://www.japheth.de/Download/Debug111.zip

1.11 [06 February 2008]
- bugfix: debugger lost debuggee's HiWord(EFL) in real-mode.
- bugfix: if debuggee set CR0 AM bit and EFL AC flag, an
exception 11h occured in the debugger.
- DEBUGX bugfix: G command was unable to set/restore breakpoints if
HiWord(EIP) wasn't clear.
- DEBUGX bugfix: U ignored HiWord of argument's offset part.
- DEBUGX bugfix: E ignored HiWord of argument's offset part.
- DEBUGX bugfix: M didn't accept first argument's offset > FFFF.
- DEBUGX bugfix: running in NTVDM and pressing Ctrl-C while debuggee
is in protected-mode no longer drops to real-mode.
- DEBUGX: RN now renders floating point register content properly
(DEBUG still just displays raw hex format for size reasons).
- DEBUGX: DI got an optional 'count' parameter.
- DEBUGX: D default segment part is checked if it is valid in pmode
and changed to value of client's DS if no.

---
MS-DOS forever!

> - bugfix: DKRNL32 initialization disabled a possibly installed FPU
> emulator.

I never have bothered too much trying to find a decent FPU emulator for my old 486 Sx, but it's a good idea, at least in theory. Most apps I use don't really need it though. And for DJGPP stuff, there's always WMemu (more accurate than the default EMU387.DXE).

However, it might be easier (more practical?) to just implement a fix for the FDIV bug on Pentiums < 120 Mhz. Granted, I don't have one, but I did search for quite a bit trying to read up on it (sounds interesting, at least):

http://www.geocities.com/snoopimeanie/fdiv_bug.zip (a few texts I collected)

> It has been fixed in the meantime and hdeditw works perfectly in DOS now

Thanks Works mostly, not perfectly (will open separate thread)

> (but won't in Win9x)

DOS > Win9x

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