home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
TopWare 18: Liquid
/
Image.iso
/
liquid
/
top1120
/
rev.doc
< prev
next >
Wrap
Text File
|
1994-04-10
|
6KB
|
139 lines
----====< VBE/AI Revision History >====----
----====< Copyright (c) 1993,1994 VESA, Inc. >====----
----====< All Rights Reserved >====----
INTRODUCTION:
This document details the revision history of the various
drivers found in the VESA VBE/AI SDK and DDK.
Revision 1.00 00.03 04/08/94
- NOTE: A bug was found in the WAVE SERVICES structures. The
wsGetLastError and wsTimerTick functions were reversed in the
structure, according to the specification. Since the specification
sets the standard, all code will have to follow the written
specification. This release of software reflects this change. All
three sample drivers have been updated to reflect this change.
vbeai.h was adjusted.
vbeai.inc was adjusted.
wave\api.asm was adjusted.
wave\pas\hardware.asm was adjusted.
wave\sb\hardware.asm was adjusted.
wave\ss\hardware.asm was adjusted.
- All driver software version numbers have been incremented to 03.
- VOLUME.EXE has more messages printed to the screen to tell the
user more about what is being performed.
- PMIDI.EXE The cheap quantization has introduced a problem in
processing the tracks out of order. This has been fixed by
changing the way the delta time is handled.
- PLAY.C has now been renamed to TESTW.EXE. The secondary program
generated from PLAY.C was REC.EXE. This is now called TESTR.EXE.
- PLAY.C is now a new program as an example of loading and playing
a list of .WAV files.
- STUB.ASM and related files have been deleted since the drivers now
can unload if first in chain.
- PMIDI.EXE now only sends ACTIVESENSING (a MIDI command) if
explicitly requested, instead of as a default state.
- PMIDI.EXE implemented a method of reducing tempo interrupt rate
down to 300 ticks or less. This is a cheap quantization.
- The WAVE drivers were not returning the correct pointer to the
completed circular buffer. They were returning just the start
of the whole buffer. The pointer now reflect the start of the
completed division.
- The WAVE drivers were not allowing for more than one block to
be registered at a time. This has now been corrected to allow
up to to 32 blocks to be registered.
- PASWAVE.COM and SBWAVE.COM had secondary bugs where, internally,
the block sizes were wrong for the "wsPlayBlock" and "wsRecordBlock"
routines.
- SSWAVE.COM had a bug that, internally, gave the wrong block
size to play using "wsPlayBlock".
Revision 1.00 00.02 03/09/94
- PASWAVE.COM and SBWAVE.COM had a problem where a starting buffer
address of 0x30008000 (linear 0x38000), with a length of 0x8000,
was considered invalid, but now fixed.
Revision 1.00 00.02 02/28/94
- SBWAVE.COM has new code that parses the environment string
BLASTER= to find the DMA, IRQ and board address.
- Incremented the software version # in all 5 drivers, to #2. The
banners in all the drivers now reflect version 1.02, meaning
VBE/AI version 1, software version #02.
- Removed dead code from PMIDI.C, just cosmetic cleanup.
- Added a switch to PMIDI.C to allow Active Sensing output
to MIDI devices to be disabled. The Turtle Beach Maui card
turns off all notes when it receives Active Sensing.
- Fixed a fatal bug in MIDI\API.ASM msPreLoadPatch. The return
instruction left one word on the stack. Corrected by changing
the return value from 10 to 12.
- Fixed a fatal bug in WAVE\API.ASM wsPauseIO. The return
instruction left one word on the stack. Corrected by changing
the return value from nothing to 2.
- INT 10, Query #2 was responding with volume info & services if the
device did not actually have volume support, but now, no longer.
- Fixed a fatal bug in WAVE\SS\HARDWARE.ASM. The volume device
check return instruction left the parameters on the stack.
Corrected by eliminating the volume info & services structures.
- Fixed a fatal bug in MIDI\OPL2\HARDWARE.ASM. The volume device
check return instruction left the parameters on the stack.
Corrected by eliminating the volume info & services structures.
- The first loaded TSR can now unload from memory if there are
no other TSRs loaded after AND the INT 10 vector matches the
first TSR's entrypoint.
- The SBWAVE.COM driver had a bug in it where auto-init commands
did not setup the buffer divsion via cmd 48.
- The MPU.COM driver info structure field, milibrary, was one byte
too long, which skewed the rest of the structure. The result was,
the mimemreq field said 29k, instead of 69h.
- The OPL2.COM driver didn't save/restore registers in the timer
tick routine, so things were returned corrupted.
- MPU.COM MIDI input callbacks had a problem in validating the
applications callback routine. MPU.COM checked the offset instead
of the segment, so zero based offsets didn't get called.
Revision 1.00 00.01 02/14/94
- 1st official release of both the SDK and DDK. All version #s
reflect VBE/AI version 1.00, and software version #1.
Revision 0.98 00.01:
- This revision was available during the VESA review process, before
the specification was ratified.