Hi Fred,
good catch ! Glad this one was spotted, it could impact many apps.
Fred Kiefer wrote:
The scanner in NSRangeFromString uses the method scanInt:, this is wrong
for 64bit systems. You could try to use scanInteger: instead and report
back. Just leave the method caching out while trying this if you have
problems with it.
you are right: the computer where it works is a 32bit system, the one
where it fails is a 64bit one.
Do you intended a "quick patch" like the one below?
It appears to fix my issue.
Index: Source/NSRange.m
===================================================================
--- Source/NSRange.m (revision 39168)
+++ Source/NSRange.m (working copy)
@@ -73,11 +73,13 @@
if ((*scanStringImp)(scanner, scanStringSel, @"{", NULL)
&& (*scanStringImp)(scanner, scanStringSel, @"location", NULL)
&& (*scanStringImp)(scanner, scanStringSel, @"=", NULL)
- && (*scanIntImp)(scanner, scanIntSel, (int*)&range.location)
+ && [scanner scanInteger: (NSInteger *)&range.location]
+ // && (*scanIntImp)(scanner, scanIntSel, (int*)&range.location)
&& (*scanStringImp)(scanner, scanStringSel, @",", NULL)
&& (*scanStringImp)(scanner, scanStringSel, @"length", NULL)
&& (*scanStringImp)(scanner, scanStringSel, @"=", NULL)
- && (*scanIntImp)(scanner, scanIntSel, (int*)&range.length)
+ && [scanner scanInteger: (NSInteger *)&range.length]
+ // && (*scanIntImp)(scanner, scanIntSel, (int*)&range.length)
&& (*scanStringImp)(scanner, scanStringSel, @"}", NULL))
return range;
else
I see if i can fix the Imp caching myself, test it also on a 32bit
machine and in case commit everything.
Riccardo
_______________________________________________
Gnustep-dev mailing list
Gnustep-dev@gnu.org
https://lists.gnu.org/mailman/listinfo/gnustep-dev
