summary.survfit(*, data.frame = TRUE) causes error with multi-state data

[steliosbl]

Hello,
I am encountering an error when trying to run summary.survfit() on a survfit.coxphms object with data.frame = TRUE. The error is always of the form:

 > Error in (function (..., row.names = NULL, check.rows = FALSE, check.names = TRUE,  : 
  arguments imply differing number of rows: .......

This error is thrown upon the execution of this line:

survival/R/summary.survfitms.R

Line 208 in 9ea6560

ndata <- data.frame(ndata)

MVE on mgus2:

# Prepare the mgus2 dataset.
mgus2$etime <- with(mgus2, ifelse(pstat == 0, futime, ptime))
event <- with(mgus2, ifelse(pstat == 0, 2 * death, 1))
mgus2$event <- factor(event, 0:2, labels = c("censor", "pcm", "death"))

# Fit Cox PH model 
msfit <- coxph(
    Surv(etime, event) ~ age + sex,
    data = mgus2, id = id
)

# The following ndata causes an error
ndata <- mgus2[1:2, c("age", "sex")]
sf <- survfit(msfit, ndata)
summary(sf, data.frame = TRUE)
# > Error in (function (..., row.names = NULL, check.rows = FALSE, check.names = TRUE,  : 
#  arguments imply differing number of rows: 1284, 1926

Investigating further, the error occurs whenever the number of rows in ndata is not a multiple of 3.
The example below tries ndata lengths from 1-30 and returns the ones that did not cause the error.

working_nrow <- unlist(lapply(1:30, function(i) {
    tryCatch({
            ndata <- mgus2[1:i, c("age", "sex")]
            sf <- survfit(msfit, ndata)
            summary(sf, data.frame = TRUE)
            return(i)
        }, error = function(e) NULL 
    )
}))
working_nrow
>  1  3  6  9 12 15 18 21 24 27 30 

Thank you for your help.