On 16/09/2012 2:33 AM, Dmitry Samersoff wrote:
Kelly,
Sorry for being later in a party.
Function receiving port number as an argument should use *in_port_t* -
rather than unsigned short - it expands to uint16_t or to uint32_t
depending to OS.
But that change in itself may not fix the problem because it is a
comparison of a 16-bit type against a value that requires at least
17-bits that the compiler complains about.
David
-Dmitry
On 2012-09-15 19:58, Kelly O'Hair wrote:
On Sep 15, 2012, at 4:10 AM, David Holmes wrote:
On 15/09/2012 11:52 AM, Kelly O'Hair wrote:
7198327: Fix mac warning error in hprof_init.c
http://cr.openjdk.java.net/~ohair/openjdk8/repo-jdk/webrev/src/share/demo/jvmti/hprof/hprof_init.c.sdiff.html
The Mac warns on the if test for the port number being> 65535
made more sense for the port to be a simple int.
Seems simpler and more correct to just do:
if (port == 0 || (int)port> 65535) {
If I can find a way to keep the cast count down, I tend to go that route.
I also know that compilers really don't pass 16bit quantities around, but 32bit
quantities,
so in my mind, passing an int is simpler and might avoid some implicit casting.
I spent too many years working on compilers :^(
Of course I could see that I should have made it an unsigned int.
So are you against this change?
-kto
David
-kto
