> From: Brian Pane [mailto:[EMAIL PROTECTED]] > Sent: 01 May 2002 09:50
> Justin Erenkrantz wrote: > > >On Tue, Apr 30, 2002 at 11:41:57PM +0200, [EMAIL PROTECTED] wrote: > > > >>-1 on anything which > >> -> shows in the output of AB of versions of AB which -can- > >> be compared different version numbers > >> > > > >Personally, I think that is your logical fallacy. I don't believe > >you can reliably compare results of ab from anything but the same > >binary under the same conditions. -- justin > > > > Agreed. Even the same binary of ab can perform so differently with > different versions of a kernel/libc/libsocket/etc that even a matching > version number in two ab test runs is no guarantee of consistent results. > > So... > > +1 on removing the version number from the default output of > ab and replacing it with a short disclaimer reminding the > user not to compare apples with oranges. Although the version number may not help in being able to compare results, it does help to track down what the state of the sourcetree was when this ab was compiled. So, +1 on the disclaimer, -0 on removing all signs of a version number. > --Brian Sander
