klibc - Nov 2003 - isspace() and other ctype.h functions

Hm, finally subscribed here, forgot to...

Anyway, I'm using klibc in udev and have added some needed functions
(like ftruncate and vsyslog).  Should I post the patches here before
commiting them to the cvs tree?

Also, it looks like ctype.h has a off-by-one bug.  isspace(' ') returns
0 right now, but if you change the function from:

__ctype_inline int isspace(int __c)
{
  return __ctypes[__c+1] & __ctype_space; */
}   
 
to
__ctype_inline int isspace(int __c)
{
  return __ctypes[__c] & __ctype_space; */
}   

things start working again :)

In looking at the __ctypes[] array, it is 0 based, so all of the +1
array changes in ctype.h should be removed.  Or am I missing something
here?

I can also check in this change if no one minds.

thanks,

greg k-h

Greg KH wrote:
> Hm, finally subscribed here, forgot to...
> 
> Anyway, I'm using klibc in udev and have added some needed functions
> (like ftruncate and vsyslog).  Should I post the patches here before
> commiting them to the cvs tree?
> 
That would be a good idea.
> Also, it looks like ctype.h has a off-by-one bug.  isspace(' ')
returns
> 0 right now, but if you change the function from:
> 
> __ctype_inline int isspace(int __c)
> {
>   return __ctypes[__c+1] & __ctype_space; */
> }   
>  
> to
> __ctype_inline int isspace(int __c)
> {
>   return __ctypes[__c] & __ctype_space; */
> }   
> 
> things start working again :)
 >
> In looking at the __ctypes[] array, it is 0 based, so all of the +1
> array changes in ctype.h should be removed.  Or am I missing something
> here?
 >

Yes, the first element should be -1 == EOF.  However, the array starts:

const unsigned char __ctypes[257] = {
   0,                            /* EOF */
   0,                            /* control character */
   0,                            /* control character */
   0,                            /* control character */
   0,                            /* control character */
   0,                            /* control character */
   0,                            /* control character */
   0,                            /* control character */
   __ctype_space,                /* BS */

... there is one too few of the middle lines; I'll check that in right away.

	-hpa

On Tue, 2003-11-11 at 23:14, Greg KH wrote:
> Anyway, I'm using klibc in udev and have added some needed functions
> (like ftruncate and vsyslog).  Should I post the patches here before
> commiting them to the cvs tree?
Nah.  If you have commit access, fire away.
> I can also check in this change if no one minds.
Sounds good.  Thanks.

	<b

On Tue, 11 Nov 2003, Greg KH wrote:
> __ctype_inline int isspace(int __c)
> {
>   return __ctypes[__c+1] & __ctype_space; */
> }   
Side remark: Is this supposed to crash if one passes an __c outside the 
allowed range (ie < -1 || >= sizeof(__ctypes) - 1)?. My understanding of 
the man page is that it should return 0, not "undefined behavior".

--Kai

Kai Germaschewski wrote:
> On Tue, 11 Nov 2003, Greg KH wrote:
> 
> 
>>__ctype_inline int isspace(int __c)
>>{
>>  return __ctypes[__c+1] & __ctype_space; */
>>}   
> 
> 
> Side remark: Is this supposed to crash if one passes an __c outside the 
> allowed range (ie < -1 || >= sizeof(__ctypes) - 1)?. My understanding
of
> the man page is that it should return 0, not "undefined
behavior".
> 
Quoth POSIX:

"The c argument is an int, the value of which the application shall 
ensure is a character representable as an unsigned char or equal to the 
value of the macro EOF.  If the argument has any other value, the 
behavior is undefined."

	-hpa