Segfault in rbind_all

[hoesler]

rbind_all applied to a list of data frames with special columns like in

dfl <- data.frame(x = I(list(1:2, 1:3, 1:4)))
rbind_all(list(dfl, dfl))

results in a segfault:

Invalid memory access of location 0x0 rip=0x10a21b5a8

 *** caught segfault ***
address 0x0, cause 'memory not mapped'

Traceback:
 1: .Call("dplyr_rbind_all", PACKAGE = "dplyr", dots)
 2: rbind_all(list(dfl, dfl))

This could be related to #459.

[yasminlucero]

I just ran into this.

---

*** caught segfault ***
address 0x20000010, cause 'memory not mapped'

Traceback:
1: .Call("dplyr_rbind_all", PACKAGE = "dplyr", dots)
2: rbind_all(data)
3: readData()
aborting ...

Segmentation fault (core dumped)

The object 'data' is a list where each element is a data.frame with 22 columns. Each data.frame in the list was produced by reading a delimited file using fread and passing this list of column classes:

class = c(rep('character', 4), rep('integer', 8), rep('numeric', 4), rep('integer', 6))

This same code runs fine on dplyr_0.2. Upgrading dplyr surfaced this issue.

I am running dplyr_0.3.0.2 and R v 3.1.2 on Ubuntu 12.04.5 LTS (GNU/Linux 3.2.0-69-generic x86_64).

[romainfrancois]

@yasminlucero please send us a reproducible example. Did you test with the development version ?

[yasminlucero]

@romainfrancois I will, as soon as I find a moment