problems with copying/editing strings

[Lafer]

Hello,

I am attempting to write a terminal interface program using PowerPC
Assembly/C, where I am using an integer-to-ASCII conversion algorithm.
Unfortunately, I have run into some difficulties that seem to deal
with the syntax/character types that am I using. Below is a segment of
my code which is intended to convert an integer to an array of ASCII
digits:

char * inttoascii(int val)
{
int i, temp1, temp2;
i = 0;
char * str;
temp1 = 1;
if(val < 0)
{
str = '-';
val = val*-1;
i++;
}
while(val > temp1)
{
temp1 = temp1*10;
}
while(temp1 >= 10)
{
temp1 = temp1/10;
temp2 = val/temp1;
str = temp2+48; *****
i++;
}
str = '\0';
return(str);
}

The ***** line is where I'm having my problems; presumably adding 48
to a string element (i.e., a character) should convert that value to
its ASCII equivalent so it can be displayed on my terminal screen, but
when debugging the problem line doesn't seem to be doing anything at
all; presumably this is an issue of syntax that you can help me with.
I have also tried '0' and '0x30' in place of 48.

Another problem I have is with copying strings into other strings, for
example I have a help message that says:

strcopy(output, "+ for add, - for subtract, ? for help");

where strcopy is the following function:

void strcopy(char b[], char a[])
{
int i;

for (i = 0; a != '\0'; i++)
b = a;
b = '\0';
}

When I try to do this conversion it gives me a machine check exception
in the simulator I'm using (SingleStep) something about error, memory
access in 0x0000000D) which it doesn't do if I use a normal strcopy
syntax where a and b are both declared as character pointers e.g. char
*a, char *b. Note that I do not have the standard library, standard
i/o or any other such libraries at my disposal for this program.

If anyone wants to see more of the code or can help me out, please
post a message here or e-mail me. Thanks!

Matt Lafer
University of Michigan

 

[Jens.Toerring]

I am attempting to write a terminal interface program using PowerPC
Assembly/C, where I am using an integer-to-ASCII conversion algorithm.
Unfortunately, I have run into some difficulties that seem to deal
with the syntax/character types that am I using. Below is a segment of
my code which is intended to convert an integer to an array of ASCII
digits:

char * inttoascii(int val)
{
int i, temp1, temp2;
i = 0;
char * str;
temp1 = 1;
if(val < 0)
{
str = '-';

You have some bad bug here: str has never been initialized, so it
probably points to some random place in memory. You have to allocate
enough memory and assign that to str before you can start to use it.
(There was another thread today about the question how long the
string can become in the worst case, i.e. how much memory you need
to allocate to be on the safe side.)

Lets assume that val is e.g. is 17. Then temp1 is now 100.

First time round temp1 is 10, thesecond time it's 1.

The first time round temp2 will be 1, the second time it's going
to be 17, which you probably don't want. You need another line
where you do

val %= temp1;

so that you get temp2 set to 7 on the second time through the loop.

str = temp2+48; *****

Making sure that this works also with non-ASCII representations is
extremely simple and doesn't cost you anything but just makes your
program easier to read. Use

str[ i ] = temp2 + '0'.

(of course only after having assigned enough memory to str).

i++;
}
str = '\0';
return(str);

Now here's another possible pitfall. If you should make str an
automatic array (instead of allocating memory for the pointer or
making it a static array) you would return a value that goes
out of scope the moment you leave the function and thus can't be
used by the caller.

The ***** line is where I'm having my problems; presumably adding 48
to a string element (i.e., a character) should convert that value to
its ASCII equivalent so it can be displayed on my terminal screen, but
when debugging the problem line doesn't seem to be doing anything at
all; presumably this is an issue of syntax that you can help me with.
I have also tried '0' and '0x30' in place of 48.

Another problem I have is with copying strings into other strings, for
example I have a help message that says:

strcopy(output, "+ for add, - for subtract, ? for help");

where strcopy is the following function:

void strcopy(char b[], char a[])
{
int i;

for (i = 0; a != '\0'; i++)
b = a;
b = '\0';
}

When I try to do this conversion it gives me a machine check exception
in the simulator I'm using (SingleStep) something about error, memory
access in 0x0000000D) which it doesn't do if I use a normal strcopy
syntax where a and b are both declared as character pointers e.g. char
*a, char *b. Note that I do not have the standard library, standard
i/o or any other such libraries at my disposal for this program.

Looks a lot as if you pass an unitialized pointer to that function.
If you do something like

char *output;
strcopy(output, "+ for add, - for subtract, ? for help");

you're in for such an error because 'output' doesn't point to any
memory you own. It might just by chance point to memory location
0x0D (but don't count on that). BTW, why don't you use the classic
construct

void strcopy( char *b, const char *a )
{
while ( *b++ = *a++ )
/* empty */ ;
}

Regards, Jens

 

[Martin Dickopp]

I am attempting to write a terminal interface program using PowerPC
Assembly/C, where I am using an integer-to-ASCII conversion algorithm.

str = temp2+48;

Making sure that this works also with non-ASCII representations is
extremely simple and doesn't cost you anything but just makes your
program easier to read. Use

str[ i ] = temp2 + '0'.

Note that the OP says he wants to write an integer-to-ASCII converter,
as opposed to an integer-to-text-in-the-host-encoding converter. If he
really means that, 48 would in fact be correct and '0' wrong on
non-ASCII systems.

I don't find it obvious what the OP means. "Terminal interface program"
could refer to a program which is supposed to run on a non-ASCII system,
but interfaces with an ASCII terminal.

 

[Jens.Toerring]

(e-mail address removed)-berlin.de writes:

I am attempting to write a terminal interface program using PowerPC
Assembly/C, where I am using an integer-to-ASCII conversion algorithm.

str = temp2+48;

Making sure that this works also with non-ASCII representations is
extremely simple and doesn't cost you anything but just makes your
program easier to read. Use

str[ i ] = temp2 + '0'.

Note that the OP says he wants to write an integer-to-ASCII converter,
as opposed to an integer-to-text-in-the-host-encoding converter. If he
really means that, 48 would in fact be correct and '0' wrong on
non-ASCII systems.

I don't find it obvious what the OP means. "Terminal interface program"
could refer to a program which is supposed to run on a non-ASCII system,
but interfaces with an ASCII terminal.

Yupp, of course you're right, I just carelessly assumed that the OP
meant int-to-string conversion when s/he wrote integer-to-ASCII.

Regards, Jens

 

[Joona I Palaste]

I am attempting to write a terminal interface program using PowerPC
Assembly/C, where I am using an integer-to-ASCII conversion algorithm.

str = temp2+48;

Making sure that this works also with non-ASCII representations is
extremely simple and doesn't cost you anything but just makes your
program easier to read. Use

str[ i ] = temp2 + '0'.

Note that the OP says he wants to write an integer-to-ASCII converter,
as opposed to an integer-to-text-in-the-host-encoding converter. If he
really means that, 48 would in fact be correct and '0' wrong on
non-ASCII systems.

I don't find it obvious what the OP means. "Terminal interface program"
could refer to a program which is supposed to run on a non-ASCII system,
but interfaces with an ASCII terminal.

He said "integer-to-ASCII" but who's to say he didn't mean "integer-
to-whatever-happens-to-be-this-platform's-encoding"? The latter would
be more useful in most common situations. Getting these muddled up can
be the result of an "all the world's ASCII" viewpoint.