You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@apr.apache.org by "William A. Rowe, Jr." <wr...@covalent.net> on 2002/02/12 02:34:14 UTC
Re: cvs commit: apr/file_io/unix mktemp.c
A question for the list - are we really working in microsecond resolutions?
(Win32 is to the nearest 10 microseconds, I'm guessing others aren't that
precise.) If so, what would you choose as a divisor, since we could certainly
pick a better divisor than USEC_PER_SEC for randomness.
> wrowe 02/02/11 17:27:09
>
> Modified: file_io/unix mktemp.c
> Log:
> I hope the clocks have sufficient resolution to make this randomization
> useful. Comments? We may need to divide, not necessarily by USEC_PER_SEC
> but perhaps by a nice binary value.
>
> Submitted by: Mladen Turk <mt...@mappingsoft.com>
>
> Revision Changes Path
> 1.21 +7 -1 apr/file_io/unix/mktemp.c
>
> --- mktemp.c 18 Jan 2002 19:16:29 -0000 1.20
> +++ mktemp.c 12 Feb 2002 01:27:09 -0000 1.21
> @@ -130,7 +136,7 @@
> apr_uint32_t randnum;
>
> if (randseed==0) {
> - randseed = time(NULL);
> + randseed = (int)apr_time_now();
> seedrandom(randseed);
> }
>
>
>
>
>
Re: cvs commit: apr/file_io/unix mktemp.c
Posted by "William A. Rowe, Jr." <wr...@covalent.net>.
++1, would you care to make it so :-?
----- Original Message -----
From: "Aaron Bannert" <aa...@clove.org>
To: "William A. Rowe, Jr." <wr...@covalent.net>
Cc: <de...@apr.apache.org>
Sent: Monday, February 11, 2002 7:44 PM
Subject: Re: cvs commit: apr/file_io/unix mktemp.c
> On Mon, Feb 11, 2002 at 07:34:14PM -0600, William Rowe wrote:
> > A question for the list - are we really working in microsecond resolutions?
> > (Win32 is to the nearest 10 microseconds, I'm guessing others aren't that
> > precise.) If so, what would you choose as a divisor, since we could certainly
> > pick a better divisor than USEC_PER_SEC for randomness.
>
> How about XOR-ing the lower and upper 32 bits?
>
> -aaron
>
Re: cvs commit: apr/file_io/unix mktemp.c
Posted by Aaron Bannert <aa...@clove.org>.
On Mon, Feb 11, 2002 at 07:34:14PM -0600, William Rowe wrote:
> A question for the list - are we really working in microsecond resolutions?
> (Win32 is to the nearest 10 microseconds, I'm guessing others aren't that
> precise.) If so, what would you choose as a divisor, since we could certainly
> pick a better divisor than USEC_PER_SEC for randomness.
How about XOR-ing the lower and upper 32 bits?
-aaron