Re: PC segmentation fault

[Riccardo Mottola]

Hi,

Wolfgang Lux wrote:
> Okay, upon a second inspection the original termination condition i < _length + 
> 1 may be right, but in that case the assignment endType = codeType(_uchar+end) is 
> wrong. In fact, you do not need this assignment and the local variable endType at 
> all. Instead the condition in the if statement should be corrected. Change
>    if ((startType != endType) || (end == _length))
> into
>    if (end == _length || startType != codeType(_uchar+end))
> so that the array access is made only within the bounds of the buffer.
Interestingly, that fixed the error, but just delayed the crash. By 
stressing a bit the parser, it ended crashing again: if you notice, the 
"start"  (start = i) will walk up... till past the end.
Since I think it is useless to assign the start to the last character, 
since it will anyway check and exit, I guarded that last startType look up.
It seems to work! I'll commit right now.


I am also checking to change many variables to NSUIntger when 
length/count is involved or at least NSInteger where -1 is used 
(Although perhaps NSNotFound could be used perhaps). I hope I won't 
break it.

Riccardo