apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <wr...@rowe-clan.net>
Subject Re: svn commit: r589911 - in /apr/apr-util/trunk/test: testdate.c testreslist.c
Date Tue, 30 Oct 2007 02:36:20 GMT
Jeff Trawick wrote:
> On 10/29/07, William A. Rowe, Jr. <wrowe@rowe-clan.net> wrote:
>> trawick@apache.org wrote:
>>> Author: trawick
>>> Date: Mon Oct 29 16:36:09 2007
>>> New Revision: 589911
>>>
>>> URL: http://svn.apache.org/viewvc?rev=589911&view=rev
>>> Log:
>>> when !APR_HAS_RANDOM, use a lame initial "random" number
>>> in test programs instead of bombing with a link failure
>>> +    guess = apr_time_now() % APR_USEC_PER_SEC;
>> To make this slightly less lame, would you take this % a number prime to
>> 2^32 for some extra flukeishness?  (can hardly call it entropy :)
> 
> Some text can be hard to parse (and not knowing what "prime to n"
> means doesn't help).  By that do you mean
> 
>   guess = apr_time_now() % APR_USEC_PER_SEC % x;
> 
> where
> 
> gcd(x, 2^32) == 1?
> 
> (or just "apr_time_now() % x" which makes more sense, or something
> completely different?)

Seed with (apr_uint32_t)(apr_time_now() % 4,294,967,291) which will
cleanly collapse the entire current time into (2^32 - 5).

Bill

Mime
View raw message