no MIDI input with dosemu ?

no MIDI input with dosemu ?

[Dave Phillips]

Greetings:

    First I must say that I'm greatly impressed by DOSemu 1.1.99. My 
primary use for it is running MIDI apps written for MS-DOS in the late 
1980s. It handles most apps well, but I recently realized that MIDI 
support (via either midid or the ln method) is for MIDI output only. 
Bummer. There are a few of us out here who would love to be able to 
record MIDI data into an app running under DOSemu, so if anyone has any 
advice on how we might achieve that goal, please let me know. TIA!

Best regards,

Dave Phillips

Re: no MIDI input with dosemu ?

[Stas Sergeev]

Hello.

Dave Phillips wrote:
> support (via either midid or the ln method) is for MIDI output only. 
> Bummer. There are a few of us out here who would love to be able to 
> record MIDI data into an app running under DOSemu
Why? Implementing midi input is trivial,
but I'm puzzled as of what's the use.
Dosemu needs a MIDI output as otherwise
you can't play games. But as for midi
recording, you can use the native Linux
programs just fine. And it will actually
work better, because under dosemu the
timing is not always accurate. So why
would anyone ever need that?

Re: no MIDI input with dosemu ?

[Dave Phillips]

Stas Sergeev wrote:

> Why? Implementing midi input is trivial,
> but I'm puzzled as of what's the use.
> Dosemu needs a MIDI output as otherwise
> you can't play games. But as for midi
> recording, you can use the native Linux
> programs just fine. And it will actually
> work better, because under dosemu the
> timing is not always accurate. So why
> would anyone ever need that? 

No, Stas, I can't use the native Linux sequencers "just fine" and their 
developers know it. Unfortunately there is no easy way for them to 
accomodate my working method (which results from amenities of the 
particular DOS sequencer I like to use). I put about nine years into 
working with a particular program (Sequencer Plus, or Seq+ for short) 
whose working method is far more amenable to me than that of any native 
Linux sequencer. I'm not much of a keyboard player, I'm a guitarist, so 
I prefer a MIDI sequencer with exceptional QWERTY keyboard data entry 
features, and none of the native Linux sequencers approach the usability 
of Seq+ in this regard. Actually I can use Seq+ under DOSemu to meet my 
needs, but a few other users have written to me about wanting MIDI input 
with Seq+ under DOSemu. Apparently they're not comfortable with the 
native Linux MIDI sequencers either.

The developers of the native Linux sequencers know of my needs quite 
well (I've been corresponding with all of them since their projects 
began). The problem lies in some of their own assumptions and some of 
the restrictions forced on them by X11. Also, Seq+ includes many MIDI 
features not found in Rosegarden-4, MusE, Jazz++, or seq24. Timing is in 
fact very good, especially when run with a Linux kernel patched for 
low-latency (I use the Planet CCRMA system). MusE and Rosegarden both 
have trouble playing some of my MIDI sequences that Seq+ handles 
perfectly under DOSemu. The plain fact is that for my working methods 
Seq+ is superior to any native Linux MIDI sequencer. I say this not to 
criticize the Linux sequencers, but merely to indicate how far they have 
to go before they reach the maturity of an application like Seq+ (which 
by 1995 had almost ten years of development).

Btw, I'm uninterested in integrated audio/MIDI sequencing, which is the 
design philosophy behind the Jazz++, MusE, and Rosegarden-4 (seq24 is 
designed more like a hardware sequencer such as the old Alesis MT8 and 
again has little value to my work methods). I'm strictly interested in 
MIDI, and it seems to me that the Linux developers are putting more time 
into audio-oriented features than into MIDI-specific features. As I 
said, they know my opinions about these matters (and we're still 
friends, incredible as it seems).

If adding MIDI input capability in DOSemu is trivial, hopefully these 
reasons will convince you that it would still be greatly appreciated by 
some of us.

Best regards,

Dave Phillips

Re: no MIDI input with dosemu ?

[Stas Sergeev]

[-- Attachment #1: Type: text/plain, Size: 731 bytes --]

Hello.

Dave Phillips wrote:
>>> If adding MIDI input capability in DOSemu is trivial, hopefully these 
>>> reasons will convince you that it would still be greatly appreciated 
>>> by some of us.
> Thanks for checking this out, Stas !
OK.
Here goes the patch for midi input. Now at
least I don't feel myself that guilty for not
contributing for the half of a year already:)

About the patch:
- Must be applied against the CVS version, as
1.1.99.1 lacks the IRQ support for MPU401
needed for vapimpu.
- Don't forget to restart configure after
applying.
- Create a ~/.dosemu/dosemu-midi_in pipe and
forward your midi data to it.
- Let me know how it goes as I cant really test
it (no midi keyboard, and I am not a musician
at all).

[-- Attachment #2: midi_in.diff --]
[-- Type: text/plain, Size: 7162 bytes --]

--- src/include/config.h.in	Thu Aug 28 00:58:37 2003
+++ src/include/config.h.in	Fri Jan  9 02:16:16 2004
@@ -42,6 +42,7 @@
 #define  DEFAULT_CONFIG_SCRIPT "builtin"      /* main configuration script */
 #define  DOSEMU_LOGLEVEL    "dosemu.loglevel" /* logging configuration file */
 #define  DOSEMU_MIDI        "dosemu-midi"     /* fifo for midi daemon */
+#define  DOSEMU_MIDI_IN     "dosemu-midi_in"  /* fifo for midi input */
 
 EXTERN char *config_script_name INIT(DEFAULT_CONFIG_SCRIPT);
 EXTERN char *config_script_path INIT(0);
@@ -63,6 +64,7 @@
 EXTERN char *dexe_load_path INIT(dosemuhdimage_default);
 EXTERN char *ipx_dos_ini_path INIT(ALTERNATE_ETC "/dos.ini");
 EXTERN char *dosemu_midi_path INIT("~/" LOCALDIR_BASE_NAME "/run/" DOSEMU_MIDI);
+EXTERN char *dosemu_midi_in_path INIT("~/" LOCALDIR_BASE_NAME "/run/" DOSEMU_MIDI_IN);
 
 #define    DOSEMU_USERS_FILE     dosemu_users_file_path
 #define    DOSEMU_LOGLEVEL_FILE  dosemu_loglevel_file_path
@@ -76,6 +78,7 @@
 #define    IPX_DOS_INI_PATH      ipx_dos_ini_path
 #define    DOSEMU_MAP_PATH       dosemu_map_file_name
 #define    DOSEMU_MIDI_PATH      dosemu_midi_path
+#define    DOSEMU_MIDI_IN_PATH   dosemu_midi_in_path
 
 #endif /* not __ASM__ */
 
--- src/include/sound.h	Sun Oct 26 19:35:57 2003
+++ src/include/sound.h	Fri Jan  9 02:01:09 2004
@@ -227,6 +227,7 @@
 
 #define QUEUE_SIZE  64
 #define Q_HOLDS(q) (q.end - q.start)
+#define Q_AVAIL(q) (QUEUE_SIZE - q.end)
 #define Q_CLEAR(q) (q.start = q.end = 0)
 #define Q_PUT(q, v) { if (q.end < QUEUE_SIZE) q.output[q.end++] = v; }
 #define Q_GET(q) ({ \
@@ -234,6 +235,12 @@
     if (!Q_HOLDS(q)) Q_CLEAR(q); \
     __ret; \
 })
+#define Q_ADD(q, buf, len) ({ \
+    int __len = len; \
+    int __to_copy = MIN(__len, Q_AVAIL(q)); \
+    memcpy(q.output + q.end, buf, __to_copy); \
+    q.end += __to_copy; \
+})
 
 typedef struct {
   uint8_t  output[QUEUE_SIZE];  /* Output Queue */
@@ -256,6 +263,8 @@
 	queue_t data;
 	/* Architecture specific procedures */
 	void (*data_write)(uint8_t data); /* process 1 MIDI byte */
+	int (*data_read)(uint8_t data[], int max_size);
+	void (*register_io_callback)(void (*io_callback)(void));
 } mpu401_info;
 
 /*
--- src/dosext/sound/sound.c	Sun Oct 26 19:35:57 2003
+++ src/dosext/sound/sound.c	Fri Jan  9 03:51:25 2004
@@ -347,7 +347,6 @@
      /* DSP 8-bit IRQ Ack - SB */
      S_printf("SB: 8-bit IRQ Ack: %x\n", SB_dsp.data);
      sb_deactivate_irq(SB_IRQ_8BIT);
-     SB_info.irq.active &= ~SB_IRQ_8BIT; /* may mean it never triggers! */
      SB_dsp.ready = 0x7f;
      if(SB_dsp.empty_state & DREQ_AT_EOI)
      {
@@ -361,7 +360,6 @@
    case 0x0F: /* 0x0F: DSP 16-bit IRQ - SB16 */
      S_printf("SB: 16-bit IRQ Ack: %x\n", SB_dsp.data);
      sb_deactivate_irq(SB_IRQ_16BIT);
-     SB_info.irq.active &= ~SB_IRQ_16BIT; /* may mean it never triggers! */
      SB_dsp.ready = 0x7f;
      if(SB_dsp.empty_state & DREQ_AT_EOI)
      {
@@ -648,7 +646,8 @@
   case 0:
     /* Read data port */
     r=Q_GET(mpu401_info.data);
-    S_printf("MPU401: Read data port = 0x%02x\n",r);
+    S_printf("MPU401: Read data port = 0x%02x, %i bytes still in queue\n",
+      r,Q_HOLDS(mpu401_info.data));
     sb_deactivate_irq(SB_IRQ_MIDI);
     break;
   case 1:
@@ -661,6 +660,17 @@
   return r;
 }
 
+static void mpu401_io_callback(void)
+{
+  char buf[QUEUE_SIZE];
+  int n;
+  n = mpu401_info.data_read(buf, QUEUE_SIZE);
+  if (n <= 0)
+    return;
+  S_printf("MPU401: Processing IO callback, %i bytes\n", n);
+  Q_ADD(mpu401_info.data, buf, n);
+  sb_activate_irq(SB_IRQ_MIDI);
+}
 
 /*
  * Main IO Routines - Write
@@ -1824,6 +1834,7 @@
 	case 1:
 		/* Write command port */
 		S_printf("MPU401: Write 0x%02x to command port\n",value);
+		Q_CLEAR(mpu401_info.data);
 		Q_PUT(mpu401_info.data, 0xfe); /* A command is sent: MPU_ACK it next time */
 		sb_activate_irq(SB_IRQ_MIDI);
 		switch (value) {
@@ -2596,7 +2607,7 @@
     SB_info.version = SB_NONE;
   }
 
-  (void) FM_driver_init();
+  FM_driver_init();
 }
 
 static void mpu401_init(void)
@@ -2627,7 +2638,9 @@
 
   Q_CLEAR(mpu401_info.data);
 
-  (void) MPU_driver_init();
+  MPU_driver_init();
+
+  mpu401_info.register_io_callback(mpu401_io_callback);
 }
 
 
@@ -2912,6 +2925,7 @@
       S_printf("SB: Untriggering scheduled IRQ\n");
       SB_dsp.empty_state &= ~IRQ_AT_EMPTY;
     }
+    SB_info.irq.active &= ~type;
     if(!(SB_info.irq.pending & type)) {
       return;
     }
--- src/base/init/parser.y.in	Sun Dec 28 06:22:09 2003
+++ src/base/init/parser.y.in	Fri Jan  9 02:14:42 2004
@@ -2215,6 +2215,7 @@
   LOCALDIR = get_dosemu_local_home();
   RUNDIR = mkdir_under(LOCALDIR, "run", 0);
   DOSEMU_MIDI_PATH = assemble_path(RUNDIR, DOSEMU_MIDI, 0);
+  DOSEMU_MIDI_IN_PATH = assemble_path(RUNDIR, DOSEMU_MIDI_IN, 0);
 }
 
 static void lax_user_checking(void)
--- src/arch/linux/dosext/sound/linux_sound.c	Sat Oct 11 16:54:05 2003
+++ src/arch/linux/dosext/sound/linux_sound.c	Fri Jan  9 02:27:26 2004
@@ -65,6 +65,7 @@
 
 /* MPU static vars */
 static int mpu_fd = -1;	             /* -1 = closed */
+static int mpu_in_fd = -1;	             /* -1 = closed */
 static boolean mpu_disabled = FALSE; /* TRUE if MIDI output disabled */
 
 static void linux_sb_dma_set_blocksize(int blocksize, int fragsize)
@@ -653,9 +654,6 @@
   SB_driver.DMA_complete        = linux_sb_dma_complete;
   SB_driver.DMA_set_blocksize   = linux_sb_dma_set_blocksize;
    
-  /* MPU-401 Functions */
-  mpu401_info.data_write = linux_mpu401_data_write;
-
   /* Miscellaneous Functions */
   SB_driver.set_speed           = linux_sb_set_speed;
   SB_driver.play_buffer         = NULL;
@@ -701,6 +699,11 @@
   return linux_sb_get_version();
 }
 
+void linux_mpu401_register_callback(void (*io_callback)(void))
+{
+  if (mpu_in_fd == -1) return;
+  add_to_io_select(mpu_in_fd, 1, io_callback);
+}
 
 void linux_mpu401_data_write(uint8_t data)
 {
@@ -724,6 +727,17 @@
 	}
 }
 
+int linux_mpu401_data_read(uint8_t data[], int max_len)
+{
+	int ret;
+	if (mpu_in_fd == -1) return 0;
+	if ((ret = read(mpu_in_fd,data,max_len)) == -1) {
+			S_printf("MPU401:[Linux] Failed to write to file 'midi' (%s)\n",
+			strerror(errno));
+	}
+	return ret;
+}
+
 
 int FM_driver_init(void)
 {
@@ -735,6 +749,14 @@
 int MPU_driver_init(void)
 {
   S_printf("MPU:[Linux] MPU Driver Initialisation Called\n");
+
+  /* MPU-401 Functions */
+  mpu401_info.data_write = linux_mpu401_data_write;
+  mpu401_info.data_read = linux_mpu401_data_read;
+  mpu401_info.register_io_callback = linux_mpu401_register_callback;
+
+  mpu_in_fd = RPT_SYSCALL(open(DOSEMU_MIDI_IN_PATH, O_RDONLY | O_NONBLOCK, 0777));
+
   mpu_disabled = FALSE;
   /* Output a MIDI byte to an external file */
   /* Added NONBLOCK to prevent hanging - Karcher */
--- src/arch/linux/dosext/sound/linux_sound.h	Mon Oct  6 22:16:11 2003
+++ src/arch/linux/dosext/sound/linux_sound.h	Fri Jan  9 01:24:20 2004
@@ -46,6 +46,8 @@
 
 void linux_sb_dma_complete(void);
 
+int linux_mpu401_data_read(uint8_t data[], int max_len);
 void linux_mpu401_data_write(uint8_t data);
+void linux_mpu401_register_callback(void (*io_callback)(void));
 
 int linux_sb_get_free_fragments(int *total, int *free, int *bytes);

Re: no MIDI input with dosemu ?

[Ryan Underwood]

[-- Attachment #1: Type: text/plain, Size: 820 bytes --]


On Thu, Jan 08, 2004 at 04:53:03PM +0300, Stas Sergeev wrote:
> Why? Implementing midi input is trivial,
> but I'm puzzled as of what's the use.
> Dosemu needs a MIDI output as otherwise
> you can't play games. But as for midi
> recording, you can use the native Linux
> programs just fine. And it will actually
> work better, because under dosemu the
> timing is not always accurate. So why
> would anyone ever need that?

On a slightly different topic, do you think DOSEMU would benefit at all
from realtime scheduling support?  Especially on slower 386/486 machines
because we don't have pentium TSC; but also to keep multitasking from
interfering too much with operation of a DOS program that doesn't expect
to be paused for any great length of time.

-- 
Ryan Underwood, <nemesis@icequake.net>

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: no MIDI input with dosemu ?

[Dave Phillips]

Greetings:

  Before anyone gets disturbed over all this, I must let you know that 
Stas has sent me a patch to test for adding MIDI input to the CVS 
DOSemu. I have compiled the sources and have slightly tested for MIDI 
input. It didn't work, but that doesn't mean Stas's code is wrong. I'm 
currently in the middle of a very heavy writing project (12,000 words 
and I'm only about 2/3 finished, so I've been stalled for a while 
(Sorry, Stas!). I'll try to make some time for more testing today.

  So, thanks to Stas we're getting there. When Seq+ gets MIDI input 
it'll be just fantastic. I agree with Steve, Linux MIDI just hasn't "got 
there" yet, though it's certainly moving right along (so far MusE has 
the most appeal for me, but I haven't had time to test the latest 
Rosegarden release).

  Little by little, the doing gets it done... ;)

Best,

== dp




steve.home@cox.net wrote:

>The problem is that MIDI is not just for games. I haven't used a better all-around MIDI sequencer than seq+ in the last 10 years. And I know other people who feel the same. I hate to say it but from a user's perspective, even a user with a strong pro-linux bias, Linux MIDI hasn't hit the mark, Jazz and Rosegarden included. If dosemu and Linux are to live up to their promise there's no reason not to have a full MIDI implementation. Some of us would really appreciate it.
>  
>
>>From: Dave Phillips <dlphilp@bright.net>
>>Date: 2004/01/08 Thu AM 10:16:00 EST
>>To: linux-msdos@vger.kernel.org
>>CC: Stas Sergeev <stsp@aknet.ru>,  steve.home@cox.net,  ametts@mindspring.com
>>Subject: Re: no MIDI input with dosemu ?
>>
>>Stas Sergeev wrote:
>>
>>    
>>
>>>Why? Implementing midi input is trivial,
>>>but I'm puzzled as of what's the use.
>>>Dosemu needs a MIDI output as otherwise
>>>you can't play games. But as for midi
>>>recording, you can use the native Linux
>>>programs just fine. And it will actually
>>>work better, because under dosemu the
>>>timing is not always accurate. So why
>>>would anyone ever need that? 
>>>      
>>>
>>No, Stas, I can't use the native Linux sequencers "just fine" and their 
>>developers know it. Unfortunately there is no easy way for them to 
>>accomodate my working method (which results from amenities of the 
>>particular DOS sequencer I like to use). I put about nine years into 
>>working with a particular program (Sequencer Plus, or Seq+ for short) 
>>whose working method is far more amenable to me than that of any native 
>>Linux sequencer. I'm not much of a keyboard player, I'm a guitarist, so 
>>I prefer a MIDI sequencer with exceptional QWERTY keyboard data entry 
>>features, and none of the native Linux sequencers approach the usability 
>>of Seq+ in this regard. Actually I can use Seq+ under DOSemu to meet my 
>>needs, but a few other users have written to me about wanting MIDI input 
>>with Seq+ under DOSemu. Apparently they're not comfortable with the 
>>native Linux MIDI sequencers either.
>>
>>The developers of the native Linux sequencers know of my needs quite 
>>well (I've been corresponding with all of them since their projects 
>>began). The problem lies in some of their own assumptions and some of 
>>the restrictions forced on them by X11. Also, Seq+ includes many MIDI 
>>features not found in Rosegarden-4, MusE, Jazz++, or seq24. Timing is in 
>>fact very good, especially when run with a Linux kernel patched for 
>>low-latency (I use the Planet CCRMA system). MusE and Rosegarden both 
>>have trouble playing some of my MIDI sequences that Seq+ handles 
>>perfectly under DOSemu. The plain fact is that for my working methods 
>>Seq+ is superior to any native Linux MIDI sequencer. I say this not to 
>>criticize the Linux sequencers, but merely to indicate how far they have 
>>to go before they reach the maturity of an application like Seq+ (which 
>>by 1995 had almost ten years of development).
>>
>>Btw, I'm uninterested in integrated audio/MIDI sequencing, which is the 
>>design philosophy behind the Jazz++, MusE, and Rosegarden-4 (seq24 is 
>>designed more like a hardware sequencer such as the old Alesis MT8 and 
>>again has little value to my work methods). I'm strictly interested in 
>>MIDI, and it seems to me that the Linux developers are putting more time 
>>into audio-oriented features than into MIDI-specific features. As I 
>>said, they know my opinions about these matters (and we're still 
>>friends, incredible as it seems).
>>
>>If adding MIDI input capability in DOSemu is trivial, hopefully these 
>>reasons will convince you that it would still be greatly appreciated by 
>>some of us.
>>
>>Best regards,
>>
>>Dave Phillips
>>
>>
>>
>>
>>    
>>
>
>
>  
>

Re: no MIDI input with dosemu ?

[Stas Sergeev]

Hello.

Dave Phillips wrote:
> I have compiled the sources and have slightly tested for MIDI 
> input. It didn't work, but that doesn't mean Stas's code is wrong.
Since I am not able to really test
the patch, it can very well have a
bug or two. But I tried to write
some crap to the midi_in pipe, and
the sequencer recoreded some noice,
so somehow it might work.
Why it doesn't work for you, is most
probably because you still run
vapimpu.com with /irq:7 option.
This is the way to cheat vapimpu
under conditions of no IRQ support
for MPU-401. This is not any more,
and you have to run vapimpu.com
without any parameters to let it
detect the proper IRQ itself. Then
it might work, hopefully.

Re: no MIDI input with dosemu ?

[Stas Sergeev]

Hello.

steve.home@cox.net wrote:
> The problem is that MIDI is not just for games. I haven't used a
> better all-around MIDI sequencer than seq+ in the last 10 years.
OK, I don't know what seq+ does, actually
I know practically nothing about that midi
stuff. But for dosemu there is only a
midi input and midi output. So as for
output, you can use it with games, sequencers,
whatever. As for input, I think (but I
may be completely wrong, since I know
nothing about midi) you can record your
midi stream to the file under linux, and
then, since dosemu already support the
midi output, you can use the dos-based
programs to edit that recorded file or
do anything else with it.
I don't know why the *recording* of a
midi file can be better under dosemu
that under linux. Even the midid, the helper
program that comes with dosemu, can record
the midi files very well.
 From that point of view I don't see the
value of implementing the midi recording
under dosemu - you should be able to use
dos-based progs even if you recorded the
song under linux.
Feel free to explain me why am I wrong
(better not CCing to linux-msdos since it
would be an off-topic here) if you would
like.
Anyway, I have implemented it already, the
patch is here, awaiting for someone's
testing:
http://marc.theaimsgroup.com/?l=linux-msdos&m=107361154412470&w=2

Re: no MIDI input with dosemu ?

[Ged Haywood]

Hi all,

On Fri, 16 Jan 2004, Stas Sergeev wrote:

> I don't know why the *recording* of a midi file can be better under
> dosemu that under linux. Even the midid, the helper program that
> comes with dosemu, can record the midi files very well.  From that
> point of view I don't see the value of implementing the midi
> recording under dosemu - you should be able to use dos-based progs
> even if you recorded the song under linux.  Feel free to explain
> [snip]
> (better not CCing to linux-msdos since it would be an off-topic here)
>  if you would like.

Please, if you don't continue the thread on the List, CC me too.
I have a friend who is very interested.  He plays a "SynthAxe"
(and it's astounding:).

73,
Ged.

Re: no MIDI input with dosemu ?

[Robert Komar]

On Fri, 16 Jan 2004, Ged Haywood wrote:

> Hi all,
>
> On Fri, 16 Jan 2004, Stas Sergeev wrote:
>
> > I don't know why the *recording* of a midi file can be better under
> > dosemu that under linux. Even the midid, the helper program that
> > comes with dosemu, can record the midi files very well.  From that
> > point of view I don't see the value of implementing the midi
> > recording under dosemu - you should be able to use dos-based progs
> > even if you recorded the song under linux.  Feel free to explain
> > [snip]
> > (better not CCing to linux-msdos since it would be an off-topic here)
> >  if you would like.
>
> Please, if you don't continue the thread on the List, CC me too.
> I have a friend who is very interested.  He plays a "SynthAxe"
> (and it's astounding:).
>
> 73,
> Ged.

Hi,
I'm also curious about how a user application like dosemu can provide
better timing than the operating system.  Please continue this thread
publicly!

Cheers,
Rob Komar

Re: no MIDI input with dosemu ?

[Dave Phillips]

Greetings:

  The short story so far is that Stas Sergeev has posted a patch that 
gives MIDI input to CVS dosemu. I don't know whether Stas's patch is 
included with the new 1.2.0 dosemu, but I imagine it can be compiled 
against it anyway (Stas, any additional instructions for your patch + 
1.2.0 ?). This patch effectively gives full-strength to DOS MIDI 
sequencers such as Sequencer Plus and Cakewalk. Also, the current dosemu 
no longer requires explicit IRQ selection for Voyetra's vapimpu.com MIDI 
interface driver; furthermore, it now accepts the last version of that 
driver (1.51) which is now freely available from Voyetra's site. I'd 
love to see if it could support the vapimqx.com driver for my Music 
Quest MQX32M, but my machine has no ISA slot for the card.

  I'll be writing a summary HOWTO in the nest week or so and will notify 
this list when it's completed.

  Vast thanks to Stas, Bart, and all the dosemu developers for the 
amazing software they've created.

  WRT timing better than the host: I'm afraid I've mis-spoke myself. Of 
course the emulator cannot provide better system timing, but the 
application's coding can certainly be better. In the case of Seq+ I've 
played MIDI files without error that will not play correctly via MusE or 
Rosegarden. I believe Seq+ may have better timing resolution, and I know 
that it was a heavily optimized program for its day. I've sent the 
offending files to the ALSA and MusE developers, but the fact remains 
that I still get better performance out of Seq+. I apologize for the 
confusion regarding my statements.

Best regards,

Dave Phillips




Robert Komar wrote:

>On Fri, 16 Jan 2004, Ged Haywood wrote:
>
>  
>
>>Hi all,
>>
>>On Fri, 16 Jan 2004, Stas Sergeev wrote:
>>
>>    
>>
>>>I don't know why the *recording* of a midi file can be better under
>>>dosemu that under linux. Even the midid, the helper program that
>>>comes with dosemu, can record the midi files very well.  From that
>>>point of view I don't see the value of implementing the midi
>>>recording under dosemu - you should be able to use dos-based progs
>>>even if you recorded the song under linux.  Feel free to explain
>>>[snip]
>>>(better not CCing to linux-msdos since it would be an off-topic here)
>>> if you would like.
>>>      
>>>
>>Please, if you don't continue the thread on the List, CC me too.
>>I have a friend who is very interested.  He plays a "SynthAxe"
>>(and it's astounding:).
>>
>>73,
>>Ged.
>>    
>>
>
>Hi,
>I'm also curious about how a user application like dosemu can provide
>better timing than the operating system.  Please continue this thread
>publicly!
>
>Cheers,
>Rob Komar
>-
>To unsubscribe from this list: send the line "unsubscribe linux-msdos" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>  
>

Re: no MIDI input with dosemu ?

[Bart Oldeman]

On Tue, 20 Jan 2004, Dave Phillips wrote:

>   The short story so far is that Stas Sergeev has posted a patch that
> gives MIDI input to CVS dosemu. I don't know whether Stas's patch is
> included with the new 1.2.0 dosemu, but I imagine it can be compiled
> against it anyway (Stas, any additional instructions for your patch +
> 1.2.0 ?).

Stas' patch applies against 1.2.0; it is included in CVS HEAD now. It'll
probably end up in 1.2.1 though, since it looks stable.

Reason is simply that for 1.2.0 I wanted to concentrate on pure bug fixes,
no new features since 1.1.99.1.

Some other features that need to be stabilized a bit more but will
probably end up in later 1.2s are:
* GPM mouse support in the console (looks stable, but not so widely
  tested)
* LFN support (has some problems in that it can venture outside the DOSEMU
  lredir "sandbox" if you carefully construct a pathname)
* bitmap fonts. Nice new feature but it's still a little broken -- you can
  get black screens which can only be solved by "mode co80"
* internationalization of filenames and direct access to VFAT short
  filenames on mounted VFAT volumes
* Clarence' direct "dosemu /path/to/dos/command/file.exe" feature.
so I'm not planning to freeze 1.2.x for the time being, except for big
code reorganizations, which really are reserved for 1.3.

Bart

Re: no MIDI input with dosemu ?

[Dave Phillips]

Bart Oldeman wrote:

>On Tue, 20 Jan 2004, Dave Phillips wrote:
>
>  
>
>>  The short story so far is that Stas Sergeev has posted a patch that
>>gives MIDI input to CVS dosemu. I don't know whether Stas's patch is
>>included with the new 1.2.0 dosemu, but I imagine it can be compiled
>>against it anyway (Stas, any additional instructions for your patch +
>>1.2.0 ?).
>>    
>>
>
>Stas' patch applies against 1.2.0; it is included in CVS HEAD now. It'll
>probably end up in 1.2.1 though, since it looks stable.
>
>Reason is simply that for 1.2.0 I wanted to concentrate on pure bug fixes,
>no new features since 1.1.99.1.
>  
>
Greetings:

  Stas's patch does apply without problems to the 1.2.0 sources. I 
tested MIDI I/O with Sequencer Plus, recording and playback both worked 
fine. Cool work, Bart... :)

Best,

++ dp

Re: no MIDI input with dosemu ?

[Justin Zygmont]

> * Clarence' direct "dosemu /path/to/dos/command/file.exe" feature.

sounds like the feature dosbox has. neat.