More about Splus 6.1 crashing with MacroMagic

["Day, Roger" <day@upci.pitt.edu>]

With Insightful's help, I learned that the reason MacroMagic crashes Splus 6.1

is that it changes control bits in the floating processor unit (but returns them upon

exiting).  It may be that other programs do the same.

To look for the problem, run

    fpu.control(flags=0,mask=0, init=F)

The return value should be 589855.

If not, try this:

    fpu.control(initialize=T)

 

Of course, that action might then cause the guilty program to crash!

 

-----Original Message-----
From: Day, Roger [mailto:day@biounix3.upci.pitt.edu]
Sent: Tuesday, March 25, 2003 8:56 PM
To: s-news@lists.biostat.wustl.edu
Subject: [S] Splus 6.1 crashing with MacroMagic

Splus 6.1 (on Win XP) was crashing violently whenever the
table() function was run.
On investigation via inspect(), the problem was in match(),
which then calls .Internal()  and cannot be further traced.
A faithful test is match(NA, 1).
When working properly, this returns NA.
When not, CRASH!   Oddly, match(NA, "a") does not lead to a crash.

A macro utility program from Iolo called Macro Magic
appears to be the culprit.  I'd previously discovered that
Macro Magic in memory causes R to crash on WinXP-
except that the crash would happen immediately rather
than while running user code.   Empirical testing shows
that it causes my Splus crashes too.  Note that over the years that I
used Splus 2000, Macro Magic did not interfere
as far as I know.

Roger Day
University of Pittsburgh Department of Biostatistics
University of Pittsburgh Cancer Institute
University of Pittsburgh Center for Biomedical Informatics Affiliated Faculty
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!SUITE 325 STERLING PLAZA
!!!!!!!201 N. CRAIG STREET
!!!!!!!PITTSBURGH, PA 15213
!!!!!!!e-mail:  day@upci.pitt.edu
!!!!!!!MY PHONE #   412-383-1573
!!!!!!!GENERAL PHONE #   412-383-1537
!!!!!!!FAX # 412-383-1535