runtime errors with iozone21

[Willem Jan Withagen]

Simon Barner wrote:

>>Maintainer is 'ports at FreeBSD.org', which means more or less that nobody 
>>feels responsable for it. I've looked under the hood, but at first sight 
>>it seem because the timing functions cannot keep upt with the fast 
>>return of the read call and time seems to be zero.
>>Now this requires a serious overhaul of iozone21 to get a more accurate 
>>timing.
>>
>>Would it not be better to just retire this port??
>>Especially zince there is a better version of iozone!
>>    
>>
>
>Hi,
>
>the reason why there a two versions of iozone can be found in iozone21's
>pkg-descr:
>
>[...]
>This is the 2.1 version of iozone.  The new 3.x+ versions of iozone have
>completely changed their testing methods, thus their output is useless
>in comparing with older statistics.
>[...]
>
>Of course, this message was added Sun Jan 9 22:51:21 2000 UTC, so maybe
>your idea of retiring the port is still right.
>  
>
Hi Simon,

I've been using iozone (all sorts of versions) for crude testing already 
quite some time. And the sign-on of iozone21 suggest 1994 which seems 
about right. Problem is that with current speed of processors the time 
measuring is no longer capable to do "the Right Time". And that makes to 
program outdated. Retiring is easy, since there is a 
better/newer/cleaner/brighter new version of this type of soap.

--WjW