[Rd] Segmentation fault using "survival" package (PR#2320)

Göran Broström gb@stat.umu.se
Mon, 25 Nov 2002 09:57:35 +0100 (CET)


On Mon, 25 Nov 2002 maechler@stat.math.ethz.ch wrote:

> >>>>> "Jerome" == Jerome Asselin <jerome@hivnet.ubc.ca>
> >>>>>     on Sat, 23 Nov 2002 00:51:51 +0100 (MET) writes:
>
>     Jerome> Full_Name: Jerome Asselin Version: 1.6.1 OS: RedHat
>     Jerome> Linux 7.2 Submission from: (NULL) (142.103.173.179)
>
>
>     Jerome> Hello,
>
>     Jerome> I get a segmentation fault when I run the following
>     Jerome> code. I wouldn't expect meaningful results because
>     Jerome> my response variable contains only missing values.
>     Jerome> However, I would expect something like a regular
>     Jerome> error (not a segmentation fault).
>
> I confirm the problem (OS: a version of Redhat Linux, 1.6.1 patched)
> running in "R -d gdb"
> shows the place :
>
> > coxph(surv ~ x, data=data)
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x4027dee4 in coxfit2 (maxiter=0x8c9cc60, nusedx=0x8c9cc40, nvarx=0x8c9cc20,
>     time=0x0, status=0x0, covar2=0x0, offset=0x0, weights=0x0, strata=0x0,
>     means=0x8c85738, Rf_beta=0x8c85710, u=0x8c856e8, imat2=0x8c856c0,
>     loglik=0x8cb3dd0, flag=0x8c9cc00, work=0x8b4ef10, eps=0x8c85698,
>     tol_chol=0x8c85670, sctest=0x8c85648) at coxfit2.c:116
> 116	coxfit2.c: No such file or directory.
> 	in coxfit2.c
>
> --
> Martin
>
> ______________________________________________
> R-devel@stat.math.ethz.ch mailing list
> http://www.stat.math.ethz.ch/mailman/listinfo/r-devel
>

The survival package obviously lacks a few basic sanity checks, some of
which I have communicated to its R maintainer Thomas L. For instance, data
sets without events used to crash R in earlier versions.

The best advise for now is to check your data before using survival.

 ---
 Göran Broström                    tel: +46 90 786 5223
 Department of Statistics          fax: +46 90 786 6614
 Umeå University                   http://www.stat.umu.se/egna/gb/
 SE-90187 Umeå, Sweden             e-mail: gb@stat.umu.se