another question about time()

[Bruno Medeiros]

Derek Parnell wrote:
> On Wed, 26 Apr 2006 22:09:05 +0200, Abby (J.P.) wrote:
> 
>> Hello, I'm trying to make a ping utility, I made a working ICMP request 
>> (I tested it with Ethereal, I get a reply from the server). But I wonder 
>> how much precise are the std.c.time.time() or the std.date.time(), 
>> because I always find a ping of 0ms (I know that it must be 1000 ticks 
>> per seconds, but I'm not sure, this really is).
>> Here is my code :
>>
>> char[] host = "www.microsoft.com"
>> InternetHost targetHost;
>> targetHost.getHostByName(host);
>>
>> InternetAddress targetAddress = new 
>> InternetAddress(targetHost.addrList[0], InternetAddress.PORT_ANY);
>>
>> Socket pingingSocket = new Socket(AddressFamily.INET, SocketType.RAW, 
>> ProtocolType.ICMP);
>>
>> //pingingSocket.connect(targetAddress);
>> //long start_time = getUTCtime();
>> int start_time;
>> std.c.time.time(&start_time);
>>
>> pingingSocket.sendTo(PING_REQUEST, targetAddress);
>>
>> char[] response;
>> pingingSocket.receive(response);
>>
>> //long end_time = getUTCtime();
>> int end_time;
>> std.c.time.time(&end_time);
>>
>> // char [] str_ping = std.string.toString((end_time - start_time));
>> char [] str_ping = std.string.toString((end_time - start_time)/2);
>> pingingSocket.close();
>> printf("ping : " ~ str_ping);
>>
>> As you can see, I tried both std.c.time and std.date functions, but 
>> str_ping is always 0. Did I done something wrong ? Should I put the 
>> start_time before the resolving getHostByName() ?
> 
> You might be better to use another undocumented module 'std.perf'.
> 
> For example ...
> 
>    import std.perf;
>    . . .
>    ProcessTimesCounter counter = new ProcessTimesCounter();
>    counter.start();
> 
>    . . .  the stuff you want to time . . .
> 
>    counter.stop();
>    writefln("Duration: %s microseconds", counter.microseconds);
> 
> 

Wouldn't clock() suffice ?

-- 
Bruno Medeiros - CS/E student
http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D