works OK on unxlngi6 though. Hmm ;-)

The same problem now with final m191.

m190 was OK on both systems.

So:

In the output from fstat:

         off_t    st_size;   /* file size, in bytes */

and it is given to read as nbytes:

     read(int d, void *buf, size_t nbytes);

size of off_t is 8 (__int64_t)
size of size_t is 4 (unsigned long)

;-)

BUT I wonder why it failed just now, in m191 and not earlier!

Created attachment 40214 [details]
Use size_t for read n_bytes argument

rt: can you please masterfix it?

This problem affects only Mac OS X/PPC because of the different sizes of these
types.

Verified on other systems as well.

Patch looks safe and I can apply it as masterfix, yes. But just for curiosity:
how can this size_t-stuff cause an error message as '[mM]akefile is not present'?

The easiest description is this one:

macmini:~/q oo$ cat q.c
#include <stdio.h>
#include <sys/types.h>

int main(int argc, char **argv) {
   off_t size = 776;
   printf("size is %d\n", size);
   printf("size is %lld\n", size);
}
macmini:~/q oo$ gcc q.c 
macmini:~/q oo$ ./a.out 
size is 0
size is 776
macmini:~/q oo$ 

-> the file "was" 0 bytes long so no arguments to makedepend were extracted from it.

Do we need this fixed for SRC680 (will be m192) only or does it also effect OOE680?

This problem was uncovered by ause060, so only SRC680 is affected.
Thanks.

Patch applied and committed as master fix for upcoming SRC680 m192.

closing