Crashproof C++

Use const at every opportunity

   Const is your most powerful anti-crash weapon. Use it at every opportunity. An additional benefit is that it makes your code self-documenting. For instance, look at this:

const char *add2strings(const char *sz1, const char *sz2);

Such a declaration guarantee's that no matter what wierd things go on within the function, it can't harm the application programmer's two strings sz1 and sz2. If any memory corruption occurs, it will be to variables within the function's scope. This, of course, greatly reduces side effect bugs. Of course, if the program has global variables, all bets are off...

Furthermore, the declaration of the return pointer as const means that the application programmer can't "reach inside" the function to corrupt its scope. For instance, if the return value is a static array of 40 characters, if the return wasn't static the application programmer could do this:

char *pchName = add2strings("Philbinoff", "James");
strcat(pchName, " is the name of the first author of the three books");
cout << "I just corrupted an internal variable of add2strings. ";
cout << "Will I see this message?/n";
cout << "Will it crash later in the program? Who knows!/n";

Fortunately, because add2strings() returns a const pointer, you'll get a compiler error on this:

char *pchName = add2strings("Philbinoff", "James");

Even if you declared pchName as a const char *, the moment you modified its contents with strcat() you'd get a compile error. The const keyword helps the programmer keep any errors localized, thus greatly reducing the likelihood of side-effect errors.


Be very careful with string manipulation

   My experience tells me the number one cause of "memory tromp" -- hangs, GPF's and other nasty and hard to find problems, is string manipulation. In the bad-old-days of C, this was unavoidable. Now, it shouldn't be a problem.


If you can, make a string class

Have you ever used strings in other languages? Isn't it nice? The penalty for a mistake is not a GPF on the other side of the program that occurs in some sites but not others. In other languages either it works or it doesn't, but you know right away. You can achieve the same level of confidence in your strings by using a string class.

There are many implementations of string classes. Stroustrup describes a simple one in chapter/section 7.11 of his book, The C++ Programming Language, Second Edition (ISBN0-201-53992-6). Your organization may make its own. Whatever you use, it should have provisions to output to and input from streams, concatenate, search/replace, etc, WITHOUT the application programmmer having to resort to pointer arithmetic. Such an implementation might use the plus sign for concatenation.


The Poor Man's String Class

If your organization hasn't made a string class, you can have many of the benefits by using C++ string streams. A string steam is a stream that writes to a string instead of a file. You can declare it without giving a maximum length, and you can keep appending strings to it. For instance:
//***** Create nametag string ******
ostrstream ost; 
ost << "My name is " << ychFname;
ost.put(0);                  //null terminate the string 
makenametag(ost.str());
//***** Also create a nametag with the last name after first *****
ost.seekp(ost.tellp()-1);    //get rid of the null termination
ost << " " << ychLname;
ost.put(0);                  //null terminate the string
makenametag(ost.str());


Avoid buffer overrun

Even if you don't use a string class, there are still ways you can intelligently handle zero terminated strings. The following is not one of them:
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strcpy(ychName, pchPerson); //CRUNCH!
cout << ychName << '/n';    //what this will output is anybody's guess.
Now handle it intelligently with strncpy instead
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strncpy(ychName, pchPerson, sizeof(ychName)-1);
ychName[sizeof(ychName)-1] = '/0';
cout << ychName << '/n';    //Magdelena Alexandra
Here's another way that won't corrupt memory:
char *pchPerson = "Magdelena Alexandra de la Romero";
char pchName = new char[strlen(pchPerson) + 1];
strcpy(pchName, pchPerson);
cout << pchName << '/n'; //Magdelena Alexandra de la Romero
//***** Of course, sooner or later pchName must be deleted *****
if(pchName)
  {
  delete(pchName)
  pchName = 0;
  }


Protect your stack

Local Variables Come Off the Stack

My biggest gripe about C++ is that local variables come off the stack. If you have a local instance of a class with a large buffer or lots of small variables, and you declare it as a local variable, you could use up your stack and cause the program to act erratically, probably at some other point in the program.
//***** This might cause a big problem *****
class bufferhandler
 {
 void zero(){*buf='/0';};
 void put(const char *pch){strncpy(buf, pch, sizeof(buf)-1);};
 const char *get(){return(buf);};
 private:
 char buf[2000];
 };
void myprocess(void)
 {
 bufferhandler bh1;           //2000+ bytes gone from the stack
 bufferhandler bh2;           //Another 2000+ gone. Will there be enough?
 bh1.put("Steve was here");
 bh2.put("Sylvia was here");
 /* . . . */
 }
One way around this (though I think it's ugly) is to use pointers:
void myprocess(void)
 {
 bufferhandler *pbh1 = new bufferhandler; //4 bytes gone for the pointer
 bufferhandler *pbh2 = new bufferhandler; //Another 4 gone. Stack is cool.
 pbh1->put("Steve was here");
 pbh2->put("Fred was here");
 /* . . . */
 //***** be sure to delete the pointers *****
 if(pbh1)
  {
  delete(pbh1);
  pbh1=0;
  }
 if(pbh2)
  {
  delete(pbh2);
  pbh2=0;
  }
 }


Don't Make This Bonehead Mistake

I occasionally make this goof. I forget the local variable is on the stack, and pass a pointer to that local variable it back as the function return. If the code using the returned pointer doesn't write to that pointer's contents, this won't crash the program. However, sometimes, depending on the memory map, the returned value will magically change as something else on the stack overwrites the now out of scope variable.

The last time I did this (1994), it created a bug that occurred once every few days :-(. It took myself and two other people a week (off and on) to narrow it to returning a local string. Now I'm EXTREMELY careful not to make this goof.

//***** BONEHEAD MISTAKE *****
char *addstrings(const char *sz1, const char *sz2)
 {
 char ychDst[100];    //life would be easier if I had made this static
 strncpy(ychDst, sz1, sizeof(ychDst) - 1);
 strncpy(ychDst + n1, sz2, sizeof(ychDst) - strlen(sz1) - 1);
 ychDst[sizeof(ychDst) - 1] = '/0';
 return(ychDst];      //BONEHEAD
 }


"Just Say No" to Memory Leak

This subject rates a page of its own.


Forget C

Forget you ever knew printf(). Use cout <<. Of course, scanf() was flirting with disaster even in the rough and ready C days. Use cin >>, or make your own input routines or classes. What about that wonderful sprintf()? Instead of this:

#include <iostream.h>
#include <stdio.h>
/* . . . */
char ychString[100];
sprintf(ychString, "%s, %s %c", ychLname, ychFname, *ychMname);
cout << ychString << '/n';
 

Try This:

#include <iostream.h>
#include <strstream.h>
/* . . . */
ostrstream ost;
ost << ychLname << ", " << ychFname << " " << *ychMname;
ost.put(0); //null terminate the string cout
cout << ost.str() << '/n';

Under construction would be an understatement for this page. Here is a bare-bones, bound to be changed, list of suggestions for taking the crash out of C++.

Use const at every opportunity

Const is your most powerful anti-crash weapon. Use it at every opportunity. An additional benefit is that it makes your code self-documenting. For instance, look at this:

const char *add2strings(const char *sz1, const char *sz2);

Such a declaration guarantee's that no matter what wierd things go on within the function, it can't harm the application programmer's two strings sz1 and sz2. If any memory corruption occurs, it will be to variables within the function's scope. This, of course, greatly reduces side effect bugs. Of course, if the program has global variables, all bets are off...

Furthermore, the declaration of the return pointer as const means that the application programmer can't "reach inside" the function to corrupt its scope. For instance, if the return value is a static array of 40 characters, if the return wasn't static the application programmer could do this:

char *pchName = add2strings("Philbinoff", "James");
strcat(pchName, " is the name of the first author of the three books");
cout << "I just corrupted an internal variable of add2strings. ";
cout << "Will I see this message?/n";
cout << "Will it crash later in the program? Who knows!/n";

Fortunately, because add2strings() returns a const pointer, you'll get a compiler error on this:

char *pchName = add2strings("Philbinoff", "James");

Even if you declared pchName as a const char *, the moment you modified its contents with strcat() you'd get a compile error. The const keyword helps the programmer keep any errors localized, thus greatly reducing the likelihood of side-effect errors.


Be very careful with string manipulation

My experience tells me the number one cause of "memory tromp" -- hangs, GPF's and other nasty and hard to find problems, is string manipulation. In the bad-old-days of C, this was unavoidable. Now, it shouldn't be a problem.


If you can, make a string class

Have you ever used strings in other languages? Isn't it nice? The penalty for a mistake is not a GPF on the other side of the program that occurs in some sites but not others. In other languages either it works or it doesn't, but you know right away. You can achieve the same level of confidence in your strings by using a string class.

There are many implementations of string classes. Stroustrup describes a simple one in chapter/section 7.11 of his book, The C++ Programming Language, Second Edition (ISBN0-201-53992-6). Your organization may make its own. Whatever you use, it should have provisions to output to and input from streams, concatenate, search/replace, etc, WITHOUT the application programmmer having to resort to pointer arithmetic. Such an implementation might use the plus sign for concatenation.


The Poor Man's String Class

If your organization hasn't made a string class, you can have many of the benefits by using C++ string streams. A string steam is a stream that writes to a string instead of a file. You can declare it without giving a maximum length, and you can keep appending strings to it. For instance:
//***** Create nametag string ******
ostrstream ost; 
ost << "My name is " << ychFname;
ost.put(0);                  //null terminate the string 
makenametag(ost.str());
//***** Also create a nametag with the last name after first *****
ost.seekp(ost.tellp()-1);    //get rid of the null termination
ost << " " << ychLname;
ost.put(0);                  //null terminate the string
makenametag(ost.str());


Avoid buffer overrun

Even if you don't use a string class, there are still ways you can intelligently handle zero terminated strings. The following is not one of them:
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strcpy(ychName, pchPerson); //CRUNCH!
cout << ychName << '/n';    //what this will output is anybody's guess.
Now handle it intelligently with strncpy instead
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strncpy(ychName, pchPerson, sizeof(ychName)-1);
ychName[sizeof(ychName)-1] = '/0';
cout << ychName << '/n';    //Magdelena Alexandra
Here's another way that won't corrupt memory:
char *pchPerson = "Magdelena Alexandra de la Romero";
char pchName = new char[strlen(pchPerson) + 1];
strcpy(pchName, pchPerson);
cout << pchName << '/n'; //Magdelena Alexandra de la Romero
//***** Of course, sooner or later pchName must be deleted *****
if(pchName)
  {
  delete(pchName)
  pchName = 0;
  }


Protect your stack

Local Variables Come Off the Stack

My biggest gripe about C++ is that local variables come off the stack. If you have a local instance of a class with a large buffer or lots of small variables, and you declare it as a local variable, you could use up your stack and cause the program to act erratically, probably at some other point in the program.
//***** This might cause a big problem *****
class bufferhandler
 {
 void zero(){*buf='/0';};
 void put(const char *pch){strncpy(buf, pch, sizeof(buf)-1);};
 const char *get(){return(buf);};
 private:
 char buf[2000];
 };
void myprocess(void)
 {
 bufferhandler bh1;           //2000+ bytes gone from the stack
 bufferhandler bh2;           //Another 2000+ gone. Will there be enough?
 bh1.put("Steve was here");
 bh2.put("Sylvia was here");
 /* . . . */
 }
One way around this (though I think it's ugly) is to use pointers:
void myprocess(void)
 {
 bufferhandler *pbh1 = new bufferhandler; //4 bytes gone for the pointer
 bufferhandler *pbh2 = new bufferhandler; //Another 4 gone. Stack is cool.
 pbh1->put("Steve was here");
 pbh2->put("Fred was here");
 /* . . . */
 //***** be sure to delete the pointers *****
 if(pbh1)
  {
  delete(pbh1);
  pbh1=0;
  }
 if(pbh2)
  {
  delete(pbh2);
  pbh2=0;
  }
 }


Don't Make This Bonehead Mistake

I occasionally make this goof. I forget the local variable is on the stack, and pass a pointer to that local variable it back as the function return. If the code using the returned pointer doesn't write to that pointer's contents, this won't crash the program. However, sometimes, depending on the memory map, the returned value will magically change as something else on the stack overwrites the now out of scope variable.

The last time I did this (1994), it created a bug that occurred once every few days :-(. It took myself and two other people a week (off and on) to narrow it to returning a local string. Now I'm EXTREMELY careful not to make this goof.

//***** BONEHEAD MISTAKE *****
char *addstrings(const char *sz1, const char *sz2)
 {
 char ychDst[100];    //life would be easier if I had made this static
 strncpy(ychDst, sz1, sizeof(ychDst) - 1);
 strncpy(ychDst + n1, sz2, sizeof(ychDst) - strlen(sz1) - 1);
 ychDst[sizeof(ychDst) - 1] = '/0';
 return(ychDst];      //BONEHEAD
 }


"Just Say No" to Memory Leak

This subject rates a page of its own.


Forget C

Forget you ever knew printf(). Use cout <<. Of course, scanf() was flirting with disaster even in the rough and ready C days. Use cin >>, or make your own input routines or classes. What about that wonderful sprintf()? Instead of this:

#include <iostream.h>
#include <stdio.h>
/* . . . */
char ychString[100];
sprintf(ychString, "%s, %s %c", ychLname, ychFname, *ychMname);
cout << ychString << '/n';
 

Try This:

#include <iostream.h>
#include <strstream.h>
/* . . . */
ostrstream ost;
ost << ychLname << ", " << ychFname << " " << *ychMname;
ost.put(0); //null terminate the string cout
cout << ost.str() << '/n';

Under construction would be an understatement for this page. Here is a bare-bones, bound to be changed, list of suggestions for taking the crash out of C++.

Use const at every opportunity

Const is your most powerful anti-crash weapon. Use it at every opportunity. An additional benefit is that it makes your code self-documenting. For instance, look at this:

const char *add2strings(const char *sz1, const char *sz2);

Such a declaration guarantee's that no matter what wierd things go on within the function, it can't harm the application programmer's two strings sz1 and sz2. If any memory corruption occurs, it will be to variables within the function's scope. This, of course, greatly reduces side effect bugs. Of course, if the program has global variables, all bets are off...

Furthermore, the declaration of the return pointer as const means that the application programmer can't "reach inside" the function to corrupt its scope. For instance, if the return value is a static array of 40 characters, if the return wasn't static the application programmer could do this:

char *pchName = add2strings("Philbinoff", "James");
strcat(pchName, " is the name of the first author of the three books");
cout << "I just corrupted an internal variable of add2strings. ";
cout << "Will I see this message?/n";
cout << "Will it crash later in the program? Who knows!/n";

Fortunately, because add2strings() returns a const pointer, you'll get a compiler error on this:

char *pchName = add2strings("Philbinoff", "James");

Even if you declared pchName as a const char *, the moment you modified its contents with strcat() you'd get a compile error. The const keyword helps the programmer keep any errors localized, thus greatly reducing the likelihood of side-effect errors.


Be very careful with string manipulation

My experience tells me the number one cause of "memory tromp" -- hangs, GPF's and other nasty and hard to find problems, is string manipulation. In the bad-old-days of C, this was unavoidable. Now, it shouldn't be a problem.


If you can, make a string class

Have you ever used strings in other languages? Isn't it nice? The penalty for a mistake is not a GPF on the other side of the program that occurs in some sites but not others. In other languages either it works or it doesn't, but you know right away. You can achieve the same level of confidence in your strings by using a string class.

There are many implementations of string classes. Stroustrup describes a simple one in chapter/section 7.11 of his book, The C++ Programming Language, Second Edition (ISBN0-201-53992-6). Your organization may make its own. Whatever you use, it should have provisions to output to and input from streams, concatenate, search/replace, etc, WITHOUT the application programmmer having to resort to pointer arithmetic. Such an implementation might use the plus sign for concatenation.


The Poor Man's String Class

If your organization hasn't made a string class, you can have many of the benefits by using C++ string streams. A string steam is a stream that writes to a string instead of a file. You can declare it without giving a maximum length, and you can keep appending strings to it. For instance:
//***** Create nametag string ******
ostrstream ost; 
ost << "My name is " << ychFname;
ost.put(0);                  //null terminate the string 
makenametag(ost.str());
//***** Also create a nametag with the last name after first *****
ost.seekp(ost.tellp()-1);    //get rid of the null termination
ost << " " << ychLname;
ost.put(0);                  //null terminate the string
makenametag(ost.str());


Avoid buffer overrun

Even if you don't use a string class, there are still ways you can intelligently handle zero terminated strings. The following is not one of them:
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strcpy(ychName, pchPerson); //CRUNCH!
cout << ychName << '/n';    //what this will output is anybody's guess.
Now handle it intelligently with strncpy instead
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strncpy(ychName, pchPerson, sizeof(ychName)-1);
ychName[sizeof(ychName)-1] = '/0';
cout << ychName << '/n';    //Magdelena Alexandra
Here's another way that won't corrupt memory:
char *pchPerson = "Magdelena Alexandra de la Romero";
char pchName = new char[strlen(pchPerson) + 1];
strcpy(pchName, pchPerson);
cout << pchName << '/n'; //Magdelena Alexandra de la Romero
//***** Of course, sooner or later pchName must be deleted *****
if(pchName)
  {
  delete(pchName)
  pchName = 0;
  }


Protect your stack

Local Variables Come Off the Stack

My biggest gripe about C++ is that local variables come off the stack. If you have a local instance of a class with a large buffer or lots of small variables, and you declare it as a local variable, you could use up your stack and cause the program to act erratically, probably at some other point in the program.
//***** This might cause a big problem *****
class bufferhandler
 {
 void zero(){*buf='/0';};
 void put(const char *pch){strncpy(buf, pch, sizeof(buf)-1);};
 const char *get(){return(buf);};
 private:
 char buf[2000];
 };
void myprocess(void)
 {
 bufferhandler bh1;           //2000+ bytes gone from the stack
 bufferhandler bh2;           //Another 2000+ gone. Will there be enough?
 bh1.put("Steve was here");
 bh2.put("Sylvia was here");
 /* . . . */
 }
One way around this (though I think it's ugly) is to use pointers:
void myprocess(void)
 {
 bufferhandler *pbh1 = new bufferhandler; //4 bytes gone for the pointer
 bufferhandler *pbh2 = new bufferhandler; //Another 4 gone. Stack is cool.
 pbh1->put("Steve was here");
 pbh2->put("Fred was here");
 /* . . . */
 //***** be sure to delete the pointers *****
 if(pbh1)
  {
  delete(pbh1);
  pbh1=0;
  }
 if(pbh2)
  {
  delete(pbh2);
  pbh2=0;
  }
 }


Don't Make This Bonehead Mistake

I occasionally make this goof. I forget the local variable is on the stack, and pass a pointer to that local variable it back as the function return. If the code using the returned pointer doesn't write to that pointer's contents, this won't crash the program. However, sometimes, depending on the memory map, the returned value will magically change as something else on the stack overwrites the now out of scope variable.

The last time I did this (1994), it created a bug that occurred once every few days :-(. It took myself and two other people a week (off and on) to narrow it to returning a local string. Now I'm EXTREMELY careful not to make this goof.

//***** BONEHEAD MISTAKE *****
char *addstrings(const char *sz1, const char *sz2)
 {
 char ychDst[100];    //life would be easier if I had made this static
 strncpy(ychDst, sz1, sizeof(ychDst) - 1);
 strncpy(ychDst + n1, sz2, sizeof(ychDst) - strlen(sz1) - 1);
 ychDst[sizeof(ychDst) - 1] = '/0';
 return(ychDst];      //BONEHEAD
 }


"Just Say No" to Memory Leak

This subject rates a page of its own.


Forget C

Forget you ever knew printf(). Use cout <<. Of course, scanf() was flirting with disaster even in the rough and ready C days. Use cin >>, or make your own input routines or classes. What about that wonderful sprintf()? Instead of this:

#include <iostream.h>
#include <stdio.h>
/* . . . */
char ychString[100];
sprintf(ychString, "%s, %s %c", ychLname, ychFname, *ychMname);
cout << ychString << '/n';
 

Try This:

#include <iostream.h>
#include <strstream.h>
/* . . . */
ostrstream ost;
ost << ychLname << ", " << ychFname << " " << *ychMname;
ost.put(0); //null terminate the string cout
cout << ost.str() << '/n';

Under construction would be an understatement for this page. Here is a bare-bones, bound to be changed, list of suggestions for taking the crash out of C++.

Use const at every opportunity

Const is your most powerful anti-crash weapon. Use it at every opportunity. An additional benefit is that it makes your code self-documenting. For instance, look at this:

const char *add2strings(const char *sz1, const char *sz2);

Such a declaration guarantee's that no matter what wierd things go on within the function, it can't harm the application programmer's two strings sz1 and sz2. If any memory corruption occurs, it will be to variables within the function's scope. This, of course, greatly reduces side effect bugs. Of course, if the program has global variables, all bets are off...

Furthermore, the declaration of the return pointer as const means that the application programmer can't "reach inside" the function to corrupt its scope. For instance, if the return value is a static array of 40 characters, if the return wasn't static the application programmer could do this:

char *pchName = add2strings("Philbinoff", "James");
strcat(pchName, " is the name of the first author of the three books");
cout << "I just corrupted an internal variable of add2strings. ";
cout << "Will I see this message?/n";
cout << "Will it crash later in the program? Who knows!/n";

Fortunately, because add2strings() returns a const pointer, you'll get a compiler error on this:

char *pchName = add2strings("Philbinoff", "James");

Even if you declared pchName as a const char *, the moment you modified its contents with strcat() you'd get a compile error. The const keyword helps the programmer keep any errors localized, thus greatly reducing the likelihood of side-effect errors.


Be very careful with string manipulation

My experience tells me the number one cause of "memory tromp" -- hangs, GPF's and other nasty and hard to find problems, is string manipulation. In the bad-old-days of C, this was unavoidable. Now, it shouldn't be a problem.


If you can, make a string class

Have you ever used strings in other languages? Isn't it nice? The penalty for a mistake is not a GPF on the other side of the program that occurs in some sites but not others. In other languages either it works or it doesn't, but you know right away. You can achieve the same level of confidence in your strings by using a string class.

There are many implementations of string classes. Stroustrup describes a simple one in chapter/section 7.11 of his book, The C++ Programming Language, Second Edition (ISBN0-201-53992-6). Your organization may make its own. Whatever you use, it should have provisions to output to and input from streams, concatenate, search/replace, etc, WITHOUT the application programmmer having to resort to pointer arithmetic. Such an implementation might use the plus sign for concatenation.


The Poor Man's String Class

If your organization hasn't made a string class, you can have many of the benefits by using C++ string streams. A string steam is a stream that writes to a string instead of a file. You can declare it without giving a maximum length, and you can keep appending strings to it. For instance:
//***** Create nametag string ******
ostrstream ost; 
ost << "My name is " << ychFname;
ost.put(0);                  //null terminate the string 
makenametag(ost.str());
//***** Also create a nametag with the last name after first *****
ost.seekp(ost.tellp()-1);    //get rid of the null termination
ost << " " << ychLname;
ost.put(0);                  //null terminate the string
makenametag(ost.str());


Avoid buffer overrun

Even if you don't use a string class, there are still ways you can intelligently handle zero terminated strings. The following is not one of them:
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strcpy(ychName, pchPerson); //CRUNCH!
cout << ychName << '/n';    //what this will output is anybody's guess.
Now handle it intelligently with strncpy instead
char *pchPerson = "Magdelena Alexandra de la Romero";
char ychName[20];
strncpy(ychName, pchPerson, sizeof(ychName)-1);
ychName[sizeof(ychName)-1] = '/0';
cout << ychName << '/n';    //Magdelena Alexandra
Here's another way that won't corrupt memory:
char *pchPerson = "Magdelena Alexandra de la Romero";
char pchName = new char[strlen(pchPerson) + 1];
strcpy(pchName, pchPerson);
cout << pchName << '/n'; //Magdelena Alexandra de la Romero
//***** Of course, sooner or later pchName must be deleted *****
if(pchName)
  {
  delete(pchName)
  pchName = 0;
  }


Protect your stack

Local Variables Come Off the Stack

My biggest gripe about C++ is that local variables come off the stack. If you have a local instance of a class with a large buffer or lots of small variables, and you declare it as a local variable, you could use up your stack and cause the program to act erratically, probably at some other point in the program.
//***** This might cause a big problem *****
class bufferhandler
 {
 void zero(){*buf='/0';};
 void put(const char *pch){strncpy(buf, pch, sizeof(buf)-1);};
 const char *get(){return(buf);};
 private:
 char buf[2000];
 };
void myprocess(void)
 {
 bufferhandler bh1;           //2000+ bytes gone from the stack
 bufferhandler bh2;           //Another 2000+ gone. Will there be enough?
 bh1.put("Steve was here");
 bh2.put("Sylvia was here");
 /* . . . */
 }
One way around this (though I think it's ugly) is to use pointers:
void myprocess(void)
 {
 bufferhandler *pbh1 = new bufferhandler; //4 bytes gone for the pointer
 bufferhandler *pbh2 = new bufferhandler; //Another 4 gone. Stack is cool.
 pbh1->put("Steve was here");
 pbh2->put("Fred was here");
 /* . . . */
 //***** be sure to delete the pointers *****
 if(pbh1)
  {
  delete(pbh1);
  pbh1=0;
  }
 if(pbh2)
  {
  delete(pbh2);
  pbh2=0;
  }
 }


Don't Make This Bonehead Mistake

I occasionally make this goof. I forget the local variable is on the stack, and pass a pointer to that local variable it back as the function return. If the code using the returned pointer doesn't write to that pointer's contents, this won't crash the program. However, sometimes, depending on the memory map, the returned value will magically change as something else on the stack overwrites the now out of scope variable.

The last time I did this (1994), it created a bug that occurred once every few days :-(. It took myself and two other people a week (off and on) to narrow it to returning a local string. Now I'm EXTREMELY careful not to make this goof.

//***** BONEHEAD MISTAKE *****
char *addstrings(const char *sz1, const char *sz2)
 {
 char ychDst[100];    //life would be easier if I had made this static
 strncpy(ychDst, sz1, sizeof(ychDst) - 1);
 strncpy(ychDst + n1, sz2, sizeof(ychDst) - strlen(sz1) - 1);
 ychDst[sizeof(ychDst) - 1] = '/0';
 return(ychDst];      //BONEHEAD
 }


"Just Say No" to Memory Leak

This subject rates a page of its own.


Forget C

Forget you ever knew printf(). Use cout <<. Of course, scanf() was flirting with disaster even in the rough and ready C days. Use cin >>, or make your own input routines or classes. What about that wonderful sprintf()? Instead of this:

#include <iostream.h>
#include <stdio.h>
/* . . . */
char ychString[100];
sprintf(ychString, "%s, %s %c", ychLname, ychFname, *ychMname);
cout << ychString << '/n';
 

Try This:

#include <iostream.h>
#include <strstream.h>
/* . . . */
ostrstream ost;
ost << ychLname << ", " << ychFname << " " << *ychMname;
ost.put(0); //null terminate the string cout
cout << ost.str() << '/n';

  • 0
    点赞
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
目标检测(Object Detection)是计算机视觉领域的一个核心问题,其主要任务是找出图像中所有感兴趣的目标(物体),并确定它们的类别和位置。以下是对目标检测的详细阐述: 一、基本概念 目标检测的任务是解决“在哪里?是什么?”的问题,即定位出图像中目标的位置并识别出目标的类别。由于各类物体具有不同的外观、形状和姿态,加上成像时光照、遮挡等因素的干扰,目标检测一直是计算机视觉领域最具挑战性的任务之一。 二、核心问题 目标检测涉及以下几个核心问题: 分类问题:判断图像中的目标属于哪个类别。 定位问题:确定目标在图像中的具体位置。 大小问题:目标可能具有不同的大小。 形状问题:目标可能具有不同的形状。 三、算法分类 基于深度学习的目标检测算法主要分为两大类: Two-stage算法:先进行区域生成(Region Proposal),生成有可能包含待检物体的预选框(Region Proposal),再通过卷积神经网络进行样本分类。常见的Two-stage算法包括R-CNN、Fast R-CNN、Faster R-CNN等。 One-stage算法:不用生成区域提议,直接在网络中提取特征来预测物体分类和位置。常见的One-stage算法包括YOLO系列(YOLOv1、YOLOv2、YOLOv3、YOLOv4、YOLOv5等)、SSD和RetinaNet等。 四、算法原理 以YOLO系列为例,YOLO将目标检测视为回归问题,将输入图像一次性划分为多个区域,直接在输出层预测边界框和类别概率。YOLO采用卷积网络来提取特征,使用全连接层来得到预测值。其网络结构通常包含多个卷积层和全连接层,通过卷积层提取图像特征,通过全连接层输出预测结果。 五、应用领域 目标检测技术已经广泛应用于各个领域,为人们的生活带来了极大的便利。以下是一些主要的应用领域: 安全监控:在商场、银行
目标检测(Object Detection)是计算机视觉领域的一个核心问题,其主要任务是找出图像中所有感兴趣的目标(物体),并确定它们的类别和位置。以下是对目标检测的详细阐述: 一、基本概念 目标检测的任务是解决“在哪里?是什么?”的问题,即定位出图像中目标的位置并识别出目标的类别。由于各类物体具有不同的外观、形状和姿态,加上成像时光照、遮挡等因素的干扰,目标检测一直是计算机视觉领域最具挑战性的任务之一。 二、核心问题 目标检测涉及以下几个核心问题: 分类问题:判断图像中的目标属于哪个类别。 定位问题:确定目标在图像中的具体位置。 大小问题:目标可能具有不同的大小。 形状问题:目标可能具有不同的形状。 三、算法分类 基于深度学习的目标检测算法主要分为两大类: Two-stage算法:先进行区域生成(Region Proposal),生成有可能包含待检物体的预选框(Region Proposal),再通过卷积神经网络进行样本分类。常见的Two-stage算法包括R-CNN、Fast R-CNN、Faster R-CNN等。 One-stage算法:不用生成区域提议,直接在网络中提取特征来预测物体分类和位置。常见的One-stage算法包括YOLO系列(YOLOv1、YOLOv2、YOLOv3、YOLOv4、YOLOv5等)、SSD和RetinaNet等。 四、算法原理 以YOLO系列为例,YOLO将目标检测视为回归问题,将输入图像一次性划分为多个区域,直接在输出层预测边界框和类别概率。YOLO采用卷积网络来提取特征,使用全连接层来得到预测值。其网络结构通常包含多个卷积层和全连接层,通过卷积层提取图像特征,通过全连接层输出预测结果。 五、应用领域 目标检测技术已经广泛应用于各个领域,为人们的生活带来了极大的便利。以下是一些主要的应用领域: 安全监控:在商场、银行
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值