[Rd] That 'make check-all' problem with the survival package

[Marc Schwartz]

> On May 16, 2015, at 6:11 AM, Hin-Tak Leung <htl10 at users.sourceforge.net> wrote:
> 
> 
> 
> ------------------------------
> On Sat, May 16, 2015 8:04 AM BST Uwe Ligges wrote:
> 
>> Not sure why this goes to R-devel. You just could have asked the 
>> maintainer. Terry Therneau is aware of it and promised he will fix it.
>> 
> 
> The quickest fix is to add cmprsk to the recommended list , and that's is an R-devel issue.


Actually, the easiest solution would be for Terry to either modify relevant code according to:

  http://cran.r-project.org/doc/manuals/r-release/R-exts.html#Suggested-packages

or perhaps better, as noted, to remove the need for cmprsk at all.

The latter raises the philosophical issue as to whether or not a Recommended package should or really needs to have any connections at all to third party CRAN packages. It seems to me that the default R distribution of “Base” and “Recommended” packages should be able to fully run in a stand alone manner without any declared external connections to other non-default packages.

The only other Recommended package that I found that has such a connection is nlme, which has a Suggests for Frank’s Hmisc. However, based upon a grep review of the package contents, there are only two code based references to Hmisc that I located:

./R/newFunc.R:264:      ## e.g. Hmisc's "labelled"
./tests/augPred_lab.R:2:if(require("Hmisc")) {

Neither of which is really needed (the first being a comment only, so benign) and of course the latter goes against the current guidance in R-exts. The second, within the if() code, uses the Hmisc label() function, which it seems to me is not really needed here.

Regards,

Marc Schwartz


> 
>> On 16.05.2015 07:22, Hin-Tak Leung wrote:
>>> 'make check-all' for current R has been showing this error in the middle
>>> for a few months now - any thought on fixing this? I think cmprsk
>>> should be either included in the recommended bundle, or
>>> the survival vignette to not depend on it. Having 'make check-all' showing
>>> glaring ERROR's for a few months seems to defeat the purpose of
>>> doing any checking at all via 'make check-all'.
>>> 
>>> FWIW, I did look at when/how the issue was introduced, but it appeared
>>> that svn://svn.r-forge.r-project.org/svnroot/survival is no longer being
>>> updated, and git://github.com/cran/survival.git only shows release jumps.
>>> Anyway, if first appears with survival 2.38-1 in February, and as the previous
>>> 2.37-7 was 13 months older, this info is of no use to anybody.
>>> I didn't write earlier as I thought the issue would go away at some point;
>>> but obviously this isn't the case after 3 months.
>>> 
>>> -----------------------------------------------
>>>  ERROR
>>> Errors in running code in vignettes:
>>> when running code in ‘compete.Rnw’
>>>   ...
>>> temp$fstat <- as.numeric(event)
>>> 
>>> temp$msex <- with(temp, 1 * (sex == "M"))
>>> 
>>> fgfit1 <- with(temp, crr(etime, fstat, cov1 = cbind(age,
>>> +     msex, mspike), failcode = 2, cencode = 1, variance = TRUE))
>>> 
>>>   When sourcing ‘compete.R’:
>>> Error: could not find function "crr"
>>> Execution halted
>>> 
>>> * checking re-building of vignette outputs ... NOTE
>>> Error in re-building vignettes:
>>>   ...
>>> Warning in coxph(Surv(futime, death) ~ group:age2 + sex + strata(group),  :
>>>   X matrix deemed to be singular; variable 23 24 25
>>> Loading required package: cmprsk
>>> Warning in library(package, lib.loc = lib.loc, character.only = TRUE, logical.return = TRUE,  :
>>>   there is no package called ‘cmprsk’
>>> 
>>> Error: processing vignette 'compete.Rnw' failed with diagnostics:
>>>  chunk 15 (label = finegray)
>>> Error in eval(expr, envir, enclos) : could not find function "crr"
>>> Execution halted
>>> 
>>> ______________________________________________
>>> R-devel at r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>> 
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel