Open swihart opened 7 years ago
So we have the warning above, here's the two errors (same one, each for 32 and 64 bit):
<snip>
> # estimate a log-gamma random effect by integrating it out
> survkit(y, censor=cens, ccov=~x, dist="loggamma", estimate=1.3,
+ integ=id, jointmode=TRUE)
At line 894 of file survkit.f
Fortran runtime error: Index '2' of dimension 1 of array 'xgam' above upper bound of 1
** running examples for arch 'x64' ... ERROR
Lines 889-894 of survkit.f:
C-----------------------------------------------------------------------
C STARTING VALUES FOR BETA(NDIM) IF COMPUTATION OF A GAMMA PARAMETER
C AS A JOINT MODE AFTER ALGEBRAIC INTEGRATION
C-----------------------------------------------------------------------
IF (NJOINT.EQ.1)
+ BETA(NDIM)= DLOG(XGAM(INTEGAM))
Test:
REAL*8 BETA(NDIMAX),XGAM(0:NCCOV+NDCOV),BETAWOC(2*MXSTRA+3)
to
real(real64) BETA(NDIMAX),XGAM(0:NCCOV+NDCOV),BETAWOC(2*MXSTRA+3)
as per http://stackoverflow.com/a/28855438/2727349. Re-run local CMD-SFT-E. Results:
ERROR, can't install. So, undo that.
* checking whether package ‘event’ can be installed ... ERROR
Installation failed.
See ‘/Users/swihartbj/github/event.Rcheck/00install.out’ for details.
* DONE
Okay, might be problem with jointmode or loggamma. Do this in survkit.Rd lines 123-126:
\dontrun{
survkit(y, censor=cens, ccov=~x, dist="loggamma", estimate=1.3,
integ=id, jointmode=TRUE)
}
re-check locally. Results:
R CMD check results
0 errors | 0 warnings | 0 notes
R CMD check succeeded
Of course locally was fine. Submit to winbuilder and see if that's enough for now to silence the errors. ...wait 10 minutes... it was not. And it appears to be anything involving "loggamma":
> # estimate a log-gamma random effect by integrating it out
> ## Not run:
> ##D survkit(y, censor=cens, ccov=~x, dist="loggamma", estimate=1.3,
> ##D integ=id, jointmode=TRUE)
> ##D
> ## End(Not run)
> # try a fixed value of the log-gamma random effect, integrating it out
> survkit(y, censor=cens, ccov=~x, dist="loggamma", estimate=1,
+ integ=id)
At line 1060 of file survkit.f
Fortran runtime error: Index '2' of dimension 1 of array 'xgam' above upper bound of 1
* checking PDF version of manual ... OK
* DONE
Status: 2 ERRORs, 1 WARNING, 1 NOTE
So I will put dontrun on everything with loggamma and resubmit.
Looking at it more closely, I think it has to do with "loggamma" and estimate being a scalar, because the loggamma above dontrun has length(estimate) = 3 and is fine. We don't run the 2nd and 3rd loggamma and resubmit to win builder:
# estimate a log-gamma random effect
survkit(y, censor=cens, ccov=~x, random=id, dist="loggamma",
estimate=c(0.1,10,0.01))
# estimate a log-gamma random effect by integrating it out
\dontrun{
survkit(y, censor=cens, ccov=~x, dist="loggamma", estimate=1.3,
integ=id, jointmode=TRUE)
# try a fixed value of the log-gamma random effect, integrating it out
survkit(y, censor=cens, ccov=~x, dist="loggamma", estimate=1,
integ=id)
}
winbuilder results:
Status: 1 WARNING, 1 NOTE
Good. Now, open another issue about the loggamma/estimate(scalar) issue and handle later. Now on to the Fortran warning.
Taking a note from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=41102, I replace all 112 instances real*8
with real(8)
in survkit.f
and do a local re-check:
* DONE
Status: OK
R CMD check results
0 errors | 0 warnings | 0 notes
R CMD check succeeded
and a winbuilder recheck:
Status: 1 NOTE
That's how we do. Travis CI also says we're good to go!
Out of all the warnings listed below, the only one that shows up CHECK is
Here's 00install: