Message-ID: <Pine.LNX.4.44.0304101130080.10470-100000@gannet.stats>
Date: 2003-04-10T10:38:29Z
From: Brian Ripley
Subject: Rprof in Windows
In-Reply-To: <Pine.SOL.4.30.0304101009280.7994-100000@fourier.math.chalmers.se>
On Thu, 10 Apr 2003, Sharon Kuhlmann-Berenzon wrote:
> I am currently using R 1.6.1 on Windows XP. I am trying to profile a
> lengthy program I have. I managed to obtain the profile twice using
> Rprof() and then summaryRprof(). But after that, all further attempts
> indicate that no events have been recorded in the .out file.
>
> > Rprof()
> > [my program]
> > Rprof(NULL)
> > summaryRprof()
> Error in summaryRprof() : no events were recorded
>
> The Rprof.out file is created, but it only contains the line on the length
> of the interval. Could there be a parameter or option that I should check
> or that might have been changed unadvertedly?
My guess is that you have a single-CPU machine: you need to leave a time
gap between Rprof() and running the code. It seems to be a bug in the
Windows runtime used in 1.6.x that is failing to start the profiling
thread: it did not happen in 1.5.1 and 1.7.0 (a week away) will contain a
workaround -- dual-CPU machines always worked correctly.
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595