Laurent Gautier wrote:
Dear list,
When calling R from C, what appears like a spurious error can be
triggered during the execution of chisq.test(x, y).
This is happening when the following conditions are met:
- x and y are anonymous C-level R vectors (they do not have a symbol),
but
This is easy to reproduce in R:
chisq.test(c(1000, 1001, 1002, 1003, 1004, 1005, 1006, 1007, 1008, 1009, 1010,
1011),
c(1000, 1001, 1002, 1003, 1004, 1005, 1006, 1007, 1008, 1009, 1010, 1011))
The simple answer is: don't do that.
It is unclear what is a reasonable label to give in such a
Thanks to you and Peter for the quick answer.
I should definitely have tried I R. It just seemed to me so unlikely
that no one ever reported that anonymous vectors would fail with
chisq.test().
Having a label that would be x or y is along the lines of something I
was thinking of, and this
On Thu, 19 Nov 2009, Laurent Gautier wrote:
Thanks to you and Peter for the quick answer.
I should definitely have tried I R. It just seemed to me so unlikely that no
one ever reported that anonymous vectors would fail with chisq.test().
Having a label that would be x or y is along the
Can you put together a minimal, self contained example, not requiring an
external system? This has the symptoms of a bug in the external code
(e.g. a missing PROTECT), but it is possible it's a bug in R.
Duncan Murdoch
On 19/11/2009 2:40 AM, Laurent Gautier wrote:
Dear list,
When calling
On 19/11/2009 6:32 AM, Duncan Murdoch wrote:
Can you put together a minimal, self contained example, not requiring an
external system? This has the symptoms of a bug in the external code
(e.g. a missing PROTECT), but it is possible it's a bug in R.
And I should have read the followups before
Dear list,
When calling R from C, what appears like a spurious error can be
triggered during the execution of chisq.test(x, y).
This is happening when the following conditions are met:
- x and y are anonymous C-level R vectors (they do not have a symbol),
but they are protected from garbage
