Time, Relativity and Distributed Companies

“The only reason for time is so that everything doesn’t happen at once.”
-Albert Einstein

One of the most difficult parts of communication within a fully distributed company is dealing with timezones. I can’t count the number of times that someone has emailed me asking “Can we have a meeting at 10 a.m.?” The obvious question here is which 10 a.m. are you asking about? Are you asking about your 10 a.m. or my 10 a.m.?  Unfortunately, it feels pedantic to ask the person to clarify what they mean, but it matters if you want everyone to show up at the same moment in time!

As I’ve mentioned before, we use Slack in place of meetings for a lot of internal communication, but we do still need to jump on phone calls from time to time. One of the skills that I’ve had to learn when trying to setup meetings is to be very explicit about the time that you mean. Think it’s easy? Try this quick little quiz. Figure out what time it is right now in the following U.S. cities, without using a map:

  1. Las Vegas
  2. Nashville
  3. New Orleans
  4. Phoenix
  5. Detroit
  6. Cleveland
  7. Louisville
  8. Pittsburgh
  9. Milwaukee
  10. Boise

I’m betting that while you might have fairly reasonable guesses, you’re not 100% positive on all of them. A few of them (like Phoenix) are particularly tricky! My recommendation is that when working with people in different timezones, it’s best if you declare the time and timezone not only for yourself but for the person you’re trying to invite. For example, let’s say that you are located in Austin, Texas and I am located in Portland, Oregon, and I want us to have a meeting at 11 in the morning. (Portland is in the Pacific Time Zone and Austin is in the Central Time Zone). I would probably ask you something like this:

     “Are you available for a call at 11:00 a.m. Pacific (1:00 p.m. Central)?”

By communicating it this way, I’m communicating to you that:

  1. We are not both in the same timezone
  2. I would like to talk to you at 1:00 in the afternoon
  3. It will still be 11:00 in the morning for me

This might seem like an obvious thing, but by being explicit, it helps remove any ambiguity and improves the overall quality of the communication. It also removes any assumption that the person you’re communicating with understands what timezones different cities are located in. The most important skill that you can develop in a distributed organization is your ability to communicate, and one of the easiest ways to improve is to become better at communicating time and timezones when talking to people.

Unintended Consequences

Disclaimer:
This post comes from my old blog back in 2004. I’m reposting it here so that I don’t lose the content. The source was hand-written HTML so the formatting probably appears a bit off.

About seven years ago I wrote some code to do Mu-Law and A-Law compression. About six years ago, I decided to publish an article along with the source code for it. You can find it here. Anyway, the other day I received an email from someone who had taken it and modified it for what he was doing. In doing so, he found a piece of misinformation that has been in my article since I originally published it. Not a big deal, and I intend to rectify the issue. However, as we chatted over email, I asked him what he was using Mu-Law/A-Law compression for. Here is a clip from our email:

> So if you don’t mind me asking, what are you working on that has 13 bit

> unsigned audio input?

Sure. I am designing a system that monitors lots of radio stations to capture and

detect Emergency Alert System activations – those ugly tones and occasional

voice messages you hear on the radio. The system has some rather incredible

shortcomings for a critical system in 2005. When the emergency management

office triggers an alert they have no way of knowing whether or not radio stations

actually broadcast the alert. Sometimes the system fails – too often. So our

system listens to the stations and sends a report back to Emergency HQ. In

most cases an exception report that shows which stations did not properly

send out the alert. So if the dam is breaking or the nuke is going critical they

can try again, use the phone, send a helicopter or something.

 

Whoa. Code that I originally wrote to compress audio in children’s games is now being used to help monitor Emergency situations. Talk about your unintended consequences!

-Jon

Translating Hardware Exceptions to C++ Exceptions

Disclaimer:
This post comes from my old blog back in 2004. I’m reposting it here so that I don’t lose the content. The source was hand-written HTML so the formatting probably appears a bit off.

No matter how careful of a programmer you are, there will always be times when a hardware
exception will occur in your code. Perhaps it was a third party component that was the
culprit. Perhaps, it was a fellow co-worker that broke something. Or maybe it was
Microsoft itself not playing fair with its documentation and/or implementations.
Whatever the case, it is often very useful to be able to capture a run-time exception
that was generated by the CPU. Sure, you can use a catch(…) to be your fail-safe, but
wouldn’t it be great to be able to convert that exception that was generated by the hardware
into a C++ exception? I created this class in order to do that very thing. In fact,
this class was the basis for my super assert that I created, because I found that I could
cause a hardware exception any time I wanted, and by using this C++ hardware exception container,
I could access each thread’s stack frame at run-time. This would eventually enable me to
perform a stack trace inside of an assert, but I will explain that more in a different
tutorial.

Anyway, I hope that this is useful to someone. I spent a while digging around in the
mire that is Microsoft’s documentation before I put this together. Perhaps this will save
someone else time in the future.

Enjoy.

-BossHogg

#ifndef HARDWARE_EXCEPTION
#define HARDWARE_EXCEPTION 1


enum HWExceptionType
{
	eIllegalMemoryAccess	= EXCEPTION_ACCESS_VIOLATION,
	eUnexpectedBreakpoint 	= EXCEPTION_BREAKPOINT,
	eDataTypeMisalignment 	= EXCEPTION_DATATYPE_MISALIGNMENT,
	eSingleStepInstruction 	= EXCEPTION_SINGLE_STEP,
	eArrayBoundsExceeded	= EXCEPTION_ARRAY_BOUNDS_EXCEEDED,
	eDenormalFloat 		= EXCEPTION_FLT_DENORMAL_OPERAND,
	eFloatDivideByZero 	= EXCEPTION_FLT_DIVIDE_BY_ZERO,
	eFloatInexactResult	= EXCEPTION_FLT_INEXACT_RESULT,
	eFloatInvalidOperation 	= EXCEPTION_FLT_INVALID_OPERATION,
	eFloatOverflow 		= EXCEPTION_FLT_OVERFLOW,
	eFloatStackCorrupted 	= EXCEPTION_FLT_STACK_CHECK,
	eFloatUnderflow 	= EXCEPTION_FLT_UNDERFLOW,
	eIntDivideByZero 	= EXCEPTION_INT_DIVIDE_BY_ZERO,
	eIntOverflow 		= EXCEPTION_INT_OVERFLOW,
	ePrivelegedInstruction 	= EXCEPTION_PRIV_INSTRUCTION,
	eUncontinuableException = EXCEPTION_NONCONTINUABLE_EXCEPTION
};


class HWException
{
     public:
          HWException(HWExceptionType aType,
                      EXCEPTION_POINTERS* pExp):
	       itsCategory(aType),
	       itsPointers(pExp),
	       itsLocation(pExp->ExceptionRecord->ExceptionAddress)
          {
	  }

	  HWExceptionType     GetCategory()  const {return itsCategory;}
	  DWORD		      GetLocation()  const {return itsLocation;}
	  EXCEPTION_POINTERS* GetSysPointer()const {return itsPointers;}

     protected:
          HWExceptionType	itsCategory;
	  DWORD			itsLocation;
	  EXCEPTION_POINTERS*	itsPointers;
};


static void HWTranslateException(unsigned int u, 
                                 EXCEPTION_POINTERS* pExp)
{
	throw HWException((HWExceptionType)u,pExp);
}

#endif



///////////////////////////////////////////////////////////////////////
Example usage:
///////////////////////////////////////////////////////////////////////

#include "windows.h"
#include "HWException.h"


int main()
{
	//Note, setting the exception translator must be done 
	//on a per thread basis.
	_set_se_translator(HWTranslateException);

	try {
		//This will cause an access violation
		char* ptr = NULL;
		*ptr = 5; 	
	}
	catch (HWException& e)
	{
		//We can now know both the type and the
		//memory location of the instruction that
		//caused the exception.  Cool!

		HWExceptionType exceptionType = e.GetCategory();
		DWORD address = e.GetLocation();
	}
	catch (...)
	{
		//If we got here, then it was some other kind
		//of C++ exception...
	}

	return 0;
}


CPU Detection Code

Disclaimer:
This post comes from my old blog back in 2004. I’m reposting it here so that I don’t lose the content. The source was hand-written HTML so the formatting probably appears a bit off.

CPU Detection Code

I dug this code up from a project that I worked on a long time ago. Unfortunately,
it is woefully out of date, especially with any of the latest P4 processors.
Also unfortunately, I don’t have a large suite of machines on which to test this,
however, I have verified a large number of these, but not all.
Also missing from the list are any AMD processors since my old companies didn’t
explicitly support AMD. Oh, well. As always, this code is to be used at your own
expense, and I guess with this particular set of code, that means a little more.
Anyway, I hope someone finds this interesting, and if you have any questions,
feel free to ask.

-BossHogg



#include “windows.h”

bool QueryCPUID();

bool QueryMMX();

bool QueryHyperThreading();

void QueryVendorString(char* string);

bool QuerySerialNumber(char* string);

void GetCPUInfoString(char* string);

unsigned long QueryCacheSize();

unsigned long QueryCPUCount();

unsigned char QueryCPUModel();

unsigned char QueryCPUFamily();

unsigned char QueryCPUStepping();

unsigned char QueryCPUType();

bool Is8086()

{

int is8086=0;

__asm {

pushf

pop ax

mov cx, ax

and ax, 0fffh

push ax

popf

pushf

pop ax

and ax, 0f000h

cmp ax, 0f000h

mov is8086, 0

jne DONE_8086_CHECK

mov is8086, 1

DONE_8086_CHECK:

};

return !!is8086;

}

bool Is80286()

{

int is80286=0;

__asm {

smsw ax

and ax, 1

or cx, 0f000h

push cx

popf

pushf

pop ax

and ax, 0f000h

mov is80286, 1

jz DONE_80286_CHECK

mov is80286, 0

DONE_80286_CHECK:

};

return !!is80286;

}

bool Is80386()

{

int is80386=0;

__asm {

pushfd

pop eax

mov ecx, eax

xor eax, 40000h

push eax

popfd

pushfd

pop eax

xor eax, ecx

mov is80386, 1

jz DONE_80386_CHECK

mov is80386, 0

DONE_80386_CHECK:

};

return !!is80386;

}

bool QueryCPUID()

{

int hasCPUID=0;

__asm

{

pushfd

pop eax

mov ecx, eax

and ecx, 0x00200000

xor eax, 0x00200000

push eax

popfd

pushfd

pop eax

and eax, 0x00200000

xor eax, ecx

mov hasCPUID, eax

};

return !!hasCPUID;

}

bool QueryMMX()

{

bool canDoMMX=false;

__asm

{

mov eax, 1 ; request for feature flags

_emit 0x0F ; CPUID on Pentiums is 0f,a2

_emit 0xA2

test edx, 0x00800000 ; is MMX technology Bit(bit 23)in feature

jz DONE_MMX_CHECK ; flags equal to 1

mov canDoMMX,1

DONE_MMX_CHECK:

};

return canDoMMX;

}

bool QueryHyperThreading()

{

unsigned int regEdx = 0;

unsigned int regEax = 0;

unsigned int vendorId[3] = {0, 0, 0};

if (!QueryCPUID())

return false;

__asm

{

xor eax, eax // call cpuid with eax = 0

cpuid // Get vendor id string

mov vendorId, ebx

mov vendorId + 4, edx

mov vendorId + 8, ecx

mov eax, 1 // call cpuid with eax = 1

cpuid

mov regEax, eax // eax contains family processor type

mov regEdx, edx // edx has info about the availability of hyper-Threading

}

if (((regEax & 0x0F00) == 0x0F00) || (regEax & 0x0F00000))

{

if (vendorId[0] == ‘uneG’ && vendorId[1] == ‘Ieni’ && vendorId[2] == ‘letn’)

{

return !!(regEdx & 0x10000000);

}

}

return false;

}

void QueryVendorString(char* string)

{

char vendorId[12];

__asm{

mov eax, 0 ; request for feature flags

_emit 0x0F ; CPUID on Pentiums is 0f,a2

_emit 0xA2

mov dword ptr vendorId, ebx

mov dword ptr vendorId[+4], edx

mov dword ptr vendorId[+8], ecx

};

memcpy(string,vendorId,12);

string[12]=0;

}

unsigned char QueryCPUStepping()

{

unsigned char _stepping;

__asm{

mov eax, 1 ; request for feature flags

_emit 0x0F ; CPUID on Pentiums is 0f,a2

_emit 0xA2

and eax, 0x0F

mov _stepping, al

};

return _stepping;

}

unsigned char QueryCPUModel()

{

unsigned char _model;

__asm{

mov eax, 1 ; request for feature flags

_emit 0x0F ; CPUID on Pentiums is 0f,a2

_emit 0xA2

shr eax, 4

and eax, 0x0F

mov _model, al

};

return _model;

}

unsigned char QueryCPUFamily()

{

unsigned char _family;

__asm{

mov eax, 1 ; request for feature flags

_emit 0x0F ; CPUID on Pentiums is 0f,a2

_emit 0xA2

shr eax, 8

and eax, 0x0F

mov _family, al

};

return _family;

}

unsigned char QueryCPUType()

{

char _type;

__asm{

mov eax, 1 ; request for feature flags

_emit 0x0F ; CPUID on Pentiums is 0f,a2

_emit 0xA2

shr eax, 12

and eax, 0x03

mov _type, al

};

return _type;

}

unsigned long QueryCPUCount()

{

SYSTEM_INFO info;

GetSystemInfo(&info;);

return info.dwNumberOfProcessors;

}

bool LookUpL2CacheSize(int infoFlag, unsigned long& cacheSize)

{

int cacheFlag = infoFlag;

cacheFlag &= 0xFF;

while (infoFlag > 0)

{

switch (cacheFlag)

{

case 0x80:

case 0x40:

cacheSize = 0;

return true;

case 0x81:

case 0x41:

cacheSize = 128;

return true;

case 0x82:

case 0x42:

cacheSize = 256;

return true;

case 0x83:

case 0x43:

cacheSize = 512;

return true;

case 0x84:

case 0x44:

cacheSize = 1024;

return true;

case 0x85:

case 0x45:

cacheSize = 2048;

return true;

}

infoFlag = infoFlag >> 8;

cacheFlag = infoFlag & 0xFF;

}

return false;

}

unsigned long QueryCacheSize()

{

char cacheRepeatCount=0;

int eaxCacheDescriptor=0;

int ebxCacheDescriptor=0;

int ecxCacheDescriptor=0;

int edxCacheDescriptor=0;

unsigned long cacheSize=0;

__asm{

//Let’s find out the amount of the cache.

//That will help us later on figure out which processor we are running…

mov eax, 2

_emit 0x0f

_emit 0xa2

mov cacheRepeatCount, al

cmp al, 1

je DONE_CACHE_CHECK

REPEAT_CACHE_DETECT:

mov eax, 2

_emit 0x0f

_emit 0xa2

dec cacheRepeatCount

jnz REPEAT_CACHE_DETECT

DONE_CACHE_CHECK:

mov eaxCacheDescriptor, eax

mov ebxCacheDescriptor, ebx

mov ecxCacheDescriptor, ecx

mov edxCacheDescriptor, edx

};

if (LookUpL2CacheSize(eaxCacheDescriptor,cacheSize))

return cacheSize;

if (LookUpL2CacheSize(ebxCacheDescriptor,cacheSize))

return cacheSize;

if (LookUpL2CacheSize(ecxCacheDescriptor,cacheSize))

return cacheSize;

if (LookUpL2CacheSize(edxCacheDescriptor,cacheSize))

return cacheSize;

return 0;

}

bool QuerySerialNumber(char* string)

{

char serialString[32];

bool isValid=false;

unsigned long one=0;

unsigned long two=0;

unsigned long three=0;

if (QueryCPUID())

{

__asm {

mov eax, 0

_emit 0x0f

_emit 0xa2

cmp eax, 3

jl DONE

mov eax, 1

_emit 0x0f

_emit 0xa2

and edx, 0x20000

jz DONE

mov one, eax

mov eax, 3

_emit 0x0f

_emit 0xa2

mov two, edx

mov three, ecx

mov isValid, 1

DONE:

};

}

if (isValid)

{

wsprintf(serialString,”%04x-%04x-%04x-%04x-%04x-%04x”,

one >> 16, one & 0x0000FFFF,

two >> 16, two & 0x0000FFFF,

three >> 16, three & 0x0000FFFF);

}

else

strcpy(serialString,”###DISABLED OR NOT PRESENT###”);

strcpy(string,serialString);

return isValid;

}

//If type==0x10, then it’s a dual processor system…

//Also, if it’s a P2, you need to check the L2 cache size to see if

//it’s a celeron or a real P2. If it’s zero, it’s a Celery.

const int cCPULookupTableSize=28;

#define MakeFingerprint(a,b,c)           ((((unsigned long)a) << 24) | \

(((unsigned long)b) << 16) | (((unsigned long)c) << 8) | ((unsigned long)0))

struct CPUFingerprint

{

//Fingerprint is (family | model | type )

unsigned long fingerprint;

char vendorID[16];

char description[64];

};

CPUFingerprint cpuLookupTable[cCPULookupTableSize]=

{

//The first four entries in the table are reserved for special processors.

{ MakeFingerprint(0x00, 0x00, 0x00), “Unknown vendor”, “Unknown processor”},

{ MakeFingerprint(0x00, 0x00, 0x00), “Unknown vendor”, “Intel8086”},

{ MakeFingerprint(0x00, 0x00, 0x00), “Unknown vendor”, “Intel80286”},

{ MakeFingerprint(0x00, 0x00, 0x00), “Unknown vendor”, “Intel80386”},

//These are the ones that support CPUID, so we start the “real”

//lookup here. Currently, this is offset number 4 in the array.

{ MakeFingerprint(0x04, 0x00, 0x00), “Unknown vendor”, “Intel486 DX”},

{ MakeFingerprint(0x04, 0x01, 0x00), “Unknown vendor”, “Intel486 DX”},

{ MakeFingerprint(0x04, 0x02, 0x00), “Unknown vendor”, “Intel486 SX”},

{ MakeFingerprint(0x04, 0x03, 0x00), “Unknown vendor”, “Intel487 or DX2”},

{ MakeFingerprint(0x04, 0x04, 0x00), “Unknown vendor”, “Intel486 SL”},

{ MakeFingerprint(0x04, 0x05, 0x00), “Unknown vendor”, “IntelSX2”},

{ MakeFingerprint(0x04, 0x07, 0x00), “Unknown vendor”, “Write-Back Enhanced IntelDX2”},

{ MakeFingerprint(0x04, 0x08, 0x00), “Unknown vendor”, “IntelDX4”},

{ MakeFingerprint(0x04, 0x08, 0x00), “Unknown vendor”, “IntelDX4 OverDrive”},

{ MakeFingerprint(0x05, 0x01, 0x00), “Unknown vendor”, “Pentium (60,66)”},

{ MakeFingerprint(0x05, 0x02, 0x00), “Unknown vendor”, “Pentium (75 – 200)”},

{ MakeFingerprint(0x05, 0x01, 0x00), “Unknown vendor”, “Pentium Overdrive (60,66)”},

{ MakeFingerprint(0x05, 0x02, 0x00), “Unknown vendor”, “Pentium Overdrive (75 – 133)”},

{ MakeFingerprint(0x05, 0x03, 0x00), “Unknown vendor”, “Pentium Overdrive for Intel486”},

{ MakeFingerprint(0x05, 0x01, 0x00), “Unknown vendor”, “Pentium Overdrive (60,66)”},

{ MakeFingerprint(0x05, 0x04, 0x00), “Unknown vendor”, “Pentium with MMX (166,200)”},

{ MakeFingerprint(0x05, 0x04, 0x00), “Unknown vendor”, “Pentium MMX Overdrive (75 – 133)”},

{ MakeFingerprint(0x06, 0x01, 0x00), “Unknown vendor”, “PentiumPro”},

{ MakeFingerprint(0x06, 0x03, 0x00), “Unknown vendor”, “PentiumII model 3”},

{ MakeFingerprint(0x06, 0x05, 0x00), “Unknown vendor”, “PentiumII model 5 or Celeron”},

{ MakeFingerprint(0x06, 0x08, 0x00), “Unknown vendor”, “PentiumII or Celeron”},

{ MakeFingerprint(0x0F, 0x00, 0x08), “Unknown vendor”, “Pentium 4” },

{ MakeFingerprint(0x0F, 0x01, 0x08), “Unknown vendor”, “Pentium 4” },

{ MakeFingerprint(0x0F, 0x02, 0x09), “Unknown vendor”, “Pentium 4” }

};

void GetCPUInfoString(char* string)

{

long tableEntry=0;

if (Is8086())

tableEntry=1;

else if (Is80286())

tableEntry=2;

else if (Is80386())

tableEntry=3;

else if (QueryCPUID())

{

unsigned char family=QueryCPUFamily();

unsigned char model=QueryCPUModel();

unsigned char type=QueryCPUType();

//unsigned char stepping=QueryCPUStepping();

unsigned long fingerprint = MakeFingerprint(family, model, type);

//Start the lookup at four to skip the reserved entries…

for (int x=4; x < cCPULookupTableSize; x++)

{

if (fingerprint == cpuLookupTable[x].fingerprint)

tableEntry=x;

}

//Fill in the vendor string…

QueryVendorString(cpuLookupTable[tableEntry].vendorID);

}

unsigned long cacheSize = QueryCacheSize();

bool hasMMX = QueryMMX();

unsigned long cpuCount=QueryCPUCount();

char serialId[64];

QuerySerialNumber(serialId);

bool hyperThreading = QueryHyperThreading();

wsprintf(string,”———————————————————\n”

“Primary cpu = %s\n”

“CPUs present: %d\n”

“Vendor string: %s\n”

“L2 cache size: %d\n”

“MMX capable: %d\n”

“Serial ID: %s\n”

“HyperThread: %d\n”

“———————————————————\n”,

cpuLookupTable[tableEntry].description,

cpuCount,

cpuLookupTable[tableEntry].vendorID,

cacheSize,

hasMMX,

serialId,

hyperThreading);

}

Disclaimer:
This post comes from my old blog back in 2004. I’m reposting it here so that I don’t lose the content. The source was hand-written HTML so the formatting probably appears a bit off.


MCI CD Control

This is the MCI control code that I wrote for my UglyCD
player
. It is fairly full featured, but if you need more, feel free to modify it to your needs. I
have neglected the error checking code, and every call to mciSendCommand should really be checked for
its return value. As usual, this code is usable at your own risk. If you have any questions, you are
always free to ask.

-BossHogg

#ifndef MCI_CONTROL
#define MCI_CONTROL 1

class MCIControl
{
     public:
          MCIControl();
          ~MCIControl();

          int     GetNumberOfTracks();
          void    Resume();
          void    Pause();
          void    Play();
          void    Stop();
          void    OpenDoor();
          void    CloseDoor();

          void    Goto(int track,int minute, int second);
		
          int     GetCurrentTrack();
          int     GetCurrentMinute();
          int     GetCurrentSecond();

     protected:
          void    Init();
          void    SetTimeFormat();
          void    GetPosition(BYTE* track,BYTE* min,BYTE* sec);

     private:
          MCIDEVICEID   itsMCIDevice;
};

#endif

#include “windows.h”

#include “MCIControl.h”

MCIControl::MCIControl() :

itsMCIDevice(0)

{

Init();

SetTimeFormat();

}

MCIControl::~MCIControl()

{

MCI_GENERIC_PARMS Info;

Info.dwCallback=0;

mciSendCommand(itsMCIDevice, MCI_CLOSE, MCI_NOTIFY, DWORD(&Info;));

}

void MCIControl::Resume()

{

MCI_PLAY_PARMS Info;

BYTE track,minute,second;

GetPosition(&track;,&minute;,&second;);

Info.dwTo=0;

Info.dwCallback=0;

Info.dwFrom = MCI_MAKE_TMSF(track,minute,second,0);

mciSendCommand(itsMCIDevice, MCI_PLAY, MCI_FROM|MCI_NOTIFY, DWORD(&Info;));

}

void MCIControl::Pause()

{

MCI_GENERIC_PARMS Info;

Info.dwCallback = 0;

mciSendCommand(itsMCIDevice, MCI_PAUSE, MCI_NOTIFY, DWORD(&Info;));

}

void MCIControl::Goto(int track,int minute, int second)

{

MCI_PLAY_PARMS Info;

Info.dwCallback=0;

Info.dwTo=0;

Info.dwFrom = MCI_MAKE_TMSF(track,minute,second,0);

mciSendCommand(itsMCIDevice, MCI_PLAY, MCI_FROM|MCI_NOTIFY, DWORD(&Info;));

}

void MCIControl::Play()

{

MCI_PLAY_PARMS Info;

Info.dwCallback=0;

Info.dwTo=0;

Info.dwFrom = MCI_MAKE_TMSF(0,0,0,0);

mciSendCommand(itsMCIDevice, MCI_PLAY, MCI_FROM|MCI_NOTIFY, DWORD(&Info;));

}

void MCIControl::Stop()

{

MCI_GENERIC_PARMS Info;

Info.dwCallback = 0;

mciSendCommand(itsMCIDevice, MCI_STOP, MCI_NOTIFY, DWORD(&Info;));

}

void MCIControl::OpenDoor()

{

MCI_STATUS_PARMS Info;

Info.dwCallback=0;

Info.dwItem=0;

Info.dwReturn=0;

Info.dwTrack=0;

mciSendCommand(itsMCIDevice, MCI_SET, MCI_SET_DOOR_OPEN, DWORD(&Info;));

}

void MCIControl::CloseDoor()

{

MCI_STATUS_PARMS Info;

Info.dwCallback=0;

Info.dwItem=0;

Info.dwReturn=0;

Info.dwTrack=0;

mciSendCommand(itsMCIDevice, MCI_SET, MCI_SET_DOOR_CLOSED, DWORD(&Info;));

}

int MCIControl::GetCurrentTrack()

{

BYTE track;

GetPosition(&track;,NULL,NULL);

return track;

}

int MCIControl::GetCurrentMinute()

{

BYTE minute;

GetPosition(NULL,&minute;,NULL);

return minute;

}

int MCIControl::GetCurrentSecond()

{

BYTE second;

GetPosition(NULL,NULL,&second;);

return second;

}

int MCIControl::GetNumberOfTracks()

{

MCI_STATUS_PARMS Info;

Info.dwCallback = 0;

Info.dwReturn = 0;

Info.dwItem = MCI_STATUS_NUMBER_OF_TRACKS;

Info.dwTrack = 0;

mciSendCommand(itsMCIDevice,MCI_STATUS,MCI_STATUS_ITEM,DWORD(&Info;));

return (int)Info.dwReturn;

}

void MCIControl::GetPosition(BYTE* track,BYTE* min,BYTE* sec)

{

MCI_STATUS_PARMS Info;

DWORD MSF;

Info.dwCallback=0;

Info.dwReturn=0;

Info.dwTrack=0;

Info.dwItem = MCI_STATUS_POSITION;

mciSendCommand(itsMCIDevice, MCI_STATUS, MCI_STATUS_ITEM, DWORD(&Info;));

MSF = Info.dwReturn;

if (track)

*track = MCI_MSF_MINUTE(MSF);

if (min)

*min = MCI_MSF_SECOND(MSF);

if (sec)

*sec = MCI_MSF_FRAME(MSF);

}

void MCIControl::SetTimeFormat()

{

MCI_SET_PARMS Info;

Info.dwCallback=0;

Info.dwTimeFormat=MCI_FORMAT_TMSF;

Info.dwAudio=0;

mciSendCommand(itsMCIDevice, MCI_SET, MCI_SET_TIME_FORMAT, DWORD(&Info;));

}

void MCIControl::Init()

{

MCI_OPEN_PARMS Info;

Info.dwCallback=0;

Info.lpstrAlias=0;

Info.lpstrElementName=0;

Info.wDeviceID=0;

Info.lpstrDeviceType=MAKEINTRESOURCE(MCI_DEVTYPE_CD_AUDIO);

mciSendCommand(0, MCI_OPEN, MCI_OPEN_TYPE|MCI_OPEN_TYPE_ID, DWORD(&Info;));

itsMCIDevice = Info.wDeviceID;

}

Mu-Law and A-Law Compression Tutorial

Disclaimer:
This post comes from my old blog back in 2004. I’m reposting it here so that I don’t lose the content. The source was hand-written HTML so the formatting probably appears a bit off.

Overview:
What are A-Law and Mu-Law compression? In the simplest terms, they are
standard forms of audio compression for 16 bit sounds. Like most audio
compression techniques, they are lossy, which means that when you expand them
back from their compressed state, they will not be exactly the same as
when you compressed them. The compression is always 2:1, meaning that
audio compressed with either of these algorithms will always be exactly half
of their original size.

Mu-Law and A-Law compression are both logarithmic forms of data compression,
and are extremely similar, as you will see in a minute. One definition of Mu-Law is



          “…a form of logarithmic data compression

          for audio data. Due to the fact that we hear logarithmically,

          sound recorded at higher levels does not require the same

          resolution as low-level sound. This allows us to disregard

          the least significant bits in high-level data. This turns

          out to resemble a logarithmic transformation. The resulting

          compression forces a 16-bit number to be represented as an 8-bit

          number.”


(www-s.ti.com/sc/psheets/spra267/spra267.pdf)

And from the comp.dsp newsgroup FAQ we also get this definition:



          Mu-law (also “u-law”) encoding is a form of logarithmic

          quantization or companding. It’s based on the observation that

          many signals are statistically more likely to be near a low

          signal level than a high signal level. Therefore, it makes

          more sense to have more quantization points near a low level

          than a high level. In a typical mu-law system, linear samples

          of 14 to 16 bits are companded to 8 bits. Most telephone

          quality codecs (including the Sparcstation’s audio codec) use

          mu-law encoded samples.



In simpler terms, this means that sound is represented as a wave, and humans
can only hear audio in the middle of the wave. We can remove data from the upper
and lower frequencies of a sound, and humans will not be able to hear a significant
difference. Both Mu-Law and A-Law take advantage of this, and are able to
compress 16-bit audio in an manner acceptable to human ears.
A-Law and Mu-Law compression appear to have been developed at around the
same time, and basically only differ by the particular logarithmic function
used to determine the translation. When we get to the work of
implementing the algorithms, you will see that the differences are nominal.
The main difference is that Mu-Law attempts to keep the top five bits of precision,
and uses a logarithmic function to determine the bottom three bits, while A-Law
compression keeps the top four bits and uses the logarithmic function to figure out
the bottom four. Both of these algorithms are used as telecommunication standards,
A-Law being used mainly in Europe, and Mu-Law being used in the United States.

DISCLAIMER:

Please understand that I am glossing over several of the details, but recognize that
the entire purpose of this document is to make two extremely useful algorithms much
more accessable to “average” programmers, like myself.

Mu-Law Compression:

As you read this explanation, remember that the purpose of the algorithm is to
compress a 16-bit source sample down to an 8-bit sample. The crux of Mu-Law
functionality is deciding which of the samples need to keep the most of their
precision. Even the “most-important” sample will still lose precision. It
simply becomes a matter of determining how much each sample loses, and minimizing
the loss on samples deemed “more important”.

To generate a compressed Mu-Law sample from an uncompressed sample, the following
algorithm is applied to the 16-bit source sample.

(Please refer to the code listing for Mu-Law compression.)

First, the algorithm first stores off the sign. It then adds in a bias value
which (due to wrapping) will cause high valued samples to lose precision.
The top five most significant bits are pulled out of the sample (which has
been previously biased). Then, the bottom three bits of the compressed byte are
generated using a small look-up table, based on the biased value of the source sample.
The 8-bit compressed sample is then finally created by logically OR’ing together
the 5 most important bits, the 3 lower bits, and the sign when applicable. The bits
are the logically NOT’ed, which I assume is for transmission reasons (although you
might not transmit your sample.)

MuLaw Compresion Code:



const int cBias = 0x84;

const int cClip = 32635;

static char MuLawCompressTable[256] =

{

     0,0,1,1,2,2,2,2,3,3,3,3,3,3,3,3,

     4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,

     5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,

     5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,

     6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,

     6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,

     6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,

     6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7

};

unsigned char LinearToMuLawSample(short sample)

{

     int sign = (sample >> 8) & 0x80;

     if (sign)

          sample = (short)-sample;

     if (sample > cClip)

          sample = cClip;

     sample = (short)(sample + cBias);

     int exponent = (int)MuLawCompressTable[(sample>>7) & 0xFF];

     int mantissa = (sample >> (exponent+3)) & 0x0F;

     int compressedByte = ~ (sign | (exponent << 4) | mantissa);

     return (unsigned char)compressedByte;

}

A-Law Compression:

As mentioned earlier, A-Law compression is extremely similar to Mu-Law compression. As you
will see, they differ primarily in the way that they keep precision. The
following is a short synopsis of the encoding algorithm, and the code example follows the
written explanation.
First, the sign is stored off. Then the code branches. If the absolute value of the source
sample is less than 256, the 16-bit sample is simply shifted down 4 bits and converted
to an 8-bit value, thus losing the top 4 bits in the process.
However, if it is more than 256, a logarithmic algorithm is applied to the sample to
determine the precision to keep. In that case, the sample is shifted down to access the
seven most significant bits of the sample. Those seven bits are then used to determine the
precision of the bottom 4 bits. Finally, the top seven bits are shifted back up four bits
to make room for the bottom 4 bits. The two are then logically OR’d together to create the
eight bit compressed sample. The sign is then applied, and the entire compressed sample
is logically XOR’d, again, I assume for transmission reasons.

A-Law Compression Code:



static char ALawCompressTable[128] =

{

     1,1,2,2,3,3,3,3,

     4,4,4,4,4,4,4,4,

     5,5,5,5,5,5,5,5,

     5,5,5,5,5,5,5,5,

     6,6,6,6,6,6,6,6,

     6,6,6,6,6,6,6,6,

     6,6,6,6,6,6,6,6,

     6,6,6,6,6,6,6,6,

     7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7,

     7,7,7,7,7,7,7,7

};

unsigned char LinearToALawSample(short sample)

{

     int sign;

     int exponent;

     int mantissa;

     unsigned char compressedByte;

     sign = ((~sample) >> 8) & 0x80;

     if (!sign)

          sample = (short)-sample;

     if (sample > cClip)

          sample = cClip;

     if (sample >= 256)

     {

          exponent = (int)ALawCompressTable[(sample >> 8) & 0x7F];

          mantissa = (sample >> (exponent + 3) ) & 0x0F;

          compressedByte = ((exponent << 4) | mantissa);

     }

     else

     {

          compressedByte = (unsigned char)(sample >> 4);

     }

     compressedByte ^= (sign ^ 0x55);

     return compressedByte;

}

Decompression:

Now, the most obvious way to decompress a compressed Mu-Law or A-Law sample would
be to reverse the algorithm. But a more efficient method exists. Consider for a moment the
fact that A-Law and Mu-Law both take a 16-bit value and crunch it down to an 8-bit value.
The reverse of that is to take an 8-bit value and turn it into a sixteen bit value. In the
graphics world, it is extremely common to represent 32 and 24 bit values with an eight bit
index into a palette table. So, why not take a page from the world of graphics and use
palettes for the Mu-Law and A-Law compression look up? Sounds good to me. In fact,
these palettes will be smaller than their 24 and 32 bit cousins because we only need to
represent 16 bit values, not 24 and 32. In a nutshell, we will create static lookup
tables to do the reverse conversion from A-Law and Mu-Law. The two differing tables
are presented below. To convert from your compressed sample back to the raw 16-bit
sample, just use your compressed sample as the index into the table, and the corresponding
value in the table is your decompressed 16-bit sample. Obviously, the downside is that
this method requires the memory overhead for the tables, but each table is only 512 bytes.
In this day and age, that’s downright cheap for the absolute fastest decompression!

Decompression Code:



static short MuLawDecompressTable[256] =

{

     -32124,-31100,-30076,-29052,-28028,-27004,-25980,-24956,

     -23932,-22908,-21884,-20860,-19836,-18812,-17788,-16764,

     -15996,-15484,-14972,-14460,-13948,-13436,-12924,-12412,

     -11900,-11388,-10876,-10364, -9852, -9340, -8828, -8316,

      -7932, -7676, -7420, -7164, -6908, -6652, -6396, -6140,

      -5884, -5628, -5372, -5116, -4860, -4604, -4348, -4092,

      -3900, -3772, -3644, -3516, -3388, -3260, -3132, -3004,

      -2876, -2748, -2620, -2492, -2364, -2236, -2108, -1980,

      -1884, -1820, -1756, -1692, -1628, -1564, -1500, -1436,

      -1372, -1308, -1244, -1180, -1116, -1052,  -988,  -924,

       -876,  -844,  -812,  -780,  -748,  -716,  -684,  -652,

       -620,  -588,  -556,  -524,  -492,  -460,  -428,  -396,

       -372,  -356,  -340,  -324,  -308,  -292,  -276,  -260,

       -244,  -228,  -212,  -196,  -180,  -164,  -148,  -132,

       -120,  -112,  -104,   -96,   -88,   -80,   -72,   -64,

        -56,   -48,   -40,   -32,   -24,   -16,    -8,     0,

      32124, 31100, 30076, 29052, 28028, 27004, 25980, 24956,

      23932, 22908, 21884, 20860, 19836, 18812, 17788, 16764,

      15996, 15484, 14972, 14460, 13948, 13436, 12924, 12412,

      11900, 11388, 10876, 10364,  9852,  9340,  8828,  8316,

       7932,  7676,  7420,  7164,  6908,  6652,  6396,  6140,

       5884,  5628,  5372,  5116,  4860,  4604,  4348,  4092,

       3900,  3772,  3644,  3516,  3388,  3260,  3132,  3004,

       2876,  2748,  2620,  2492,  2364,  2236,  2108,  1980,

       1884,  1820,  1756,  1692,  1628,  1564,  1500,  1436,

       1372,  1308,  1244,  1180,  1116,  1052,   988,   924,

        876,   844,   812,   780,   748,   716,   684,   652,

        620,   588,   556,   524,   492,   460,   428,   396,

        372,   356,   340,   324,   308,   292,   276,   260,

        244,   228,   212,   196,   180,   164,   148,   132,

        120,   112,   104,    96,    88,    80,    72,    64,

         56,    48,    40,    32,    24,    16,     8,     0

};



static short ALawDecompressTable[256] =

{

     -5504, -5248, -6016, -5760, -4480, -4224, -4992, -4736,

     -7552, -7296, -8064, -7808, -6528, -6272, -7040, -6784,

     -2752, -2624, -3008, -2880, -2240, -2112, -2496, -2368,

     -3776, -3648, -4032, -3904, -3264, -3136, -3520, -3392,

     -22016,-20992,-24064,-23040,-17920,-16896,-19968,-18944,

     -30208,-29184,-32256,-31232,-26112,-25088,-28160,-27136,

     -11008,-10496,-12032,-11520,-8960, -8448, -9984, -9472,

     -15104,-14592,-16128,-15616,-13056,-12544,-14080,-13568,

     -344,  -328,  -376,  -360,  -280,  -264,  -312,  -296,

     -472,  -456,  -504,  -488,  -408,  -392,  -440,  -424,

     -88,   -72,   -120,  -104,  -24,   -8,    -56,   -40,

     -216,  -200,  -248,  -232,  -152,  -136,  -184,  -168,

     -1376, -1312, -1504, -1440, -1120, -1056, -1248, -1184,

     -1888, -1824, -2016, -1952, -1632, -1568, -1760, -1696,

     -688,  -656,  -752,  -720,  -560,  -528,  -624,  -592,

     -944,  -912,  -1008, -976,  -816,  -784,  -880,  -848,

      5504,  5248,  6016,  5760,  4480,  4224,  4992,  4736,

      7552,  7296,  8064,  7808,  6528,  6272,  7040,  6784,

      2752,  2624,  3008,  2880,  2240,  2112,  2496,  2368,

      3776,  3648,  4032,  3904,  3264,  3136,  3520,  3392,

      22016, 20992, 24064, 23040, 17920, 16896, 19968, 18944,

      30208, 29184, 32256, 31232, 26112, 25088, 28160, 27136,

      11008, 10496, 12032, 11520, 8960,  8448,  9984,  9472,

      15104, 14592, 16128, 15616, 13056, 12544, 14080, 13568,

      344,   328,   376,   360,   280,   264,   312,   296,

      472,   456,   504,   488,   408,   392,   440,   424,

      88,    72,   120,   104,    24,     8,    56,    40,

      216,   200,   248,   232,   152,   136,   184,   168,

      1376,  1312,  1504,  1440,  1120,  1056,  1248,  1184,

      1888,  1824,  2016,  1952,  1632,  1568,  1760,  1696,

      688,   656,   752,   720,   560,   528,   624,   592,

      944,   912,  1008,   976,   816,   784,   880,   848

};

Some Not-So-Random Oregon Statistics

Recently, I heard a fascinating interview with Larry Krasner, the DA for Philadelphia. It really challenged some of my thoughts on incarceration and inspired me to do a little research, locally. Here are some interesting statistics that I found for Oregon:

  • 2017-2019 Dept of Corrections Budget: $1.76B ($880m annually) Source
  • Average Daily Prison Population: 14,835 Source
  • Average cost per inmate: $59,319 ($880m / 14,835)

So who is in our prison system? Here are some other things I found: Source

  • White 74.5% (vs 76.4% of normal state population)
  • Hispanic 12.1% (vs 12.8% of normal state population)
  • Black 9% (vs 2.1% of normal state population)
  • 62.7% are over the age of 30
  • Male 81%
  • Female 19%

Top 5 incarceration rates by county: Source

  1. Sherman County, 2.83 / 1000
  2. Marion County, 2.14 / 1000
  3. Jefferson County, 1.96 / 1000
  4. Linn County, 1.89 / 1000
  5. Clatsop County, 1.86 / 1000

And what did they do to get there? Source

  1. Drugs 20%
  2. Assault 13.4%
  3. Other 12.6%
  4. Theft 9.3%
  5. Burglary 8.9%

This also made me curious relative to education in Oregon. Here are a few statistics I found:

  • Average Teacher Salary: $59,204 Source
  • Total Number of Teachers: 22,357 Source
  • Median Class Size: 25 Source

So here are some of my random thoughts after doing this research:

  • It’s fascinating that every incarcerated inmate almost exactly equals a full time teacher. (These costs don’t even include local and municipal jail costs.)
  • African American Oregonians are clearly incarcerated at a much higher rate than either caucasian or hispanic Oregonians.
  • It’s ALARMING that 12.6% of the prison population is incarcerated for offenses labeled as “other”. Take a look at the source for the incarceration cause breakdown and you can see how granular it gets which makes the “other” classification that much more troublesome. (Forgery weighs in at a whopping 0.4%)
  • The likelihood of incarceration has nothing to do with the population density that you live in. Only one of the top 5 counties in the per capita list would be considered urban.

Does any of this mean anything? I’m not sure. I’ll definitely be thinking about it for a while. In the alarmist era we currently find ourselves in, I find it helpful to have some actual data behind to fall back on. As such, I’ve included links to all of my sources in case anyone else feels so inspired to do some data spelunking. If you do, please share what you find!