You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: NEWS.md
+2Lines changed: 2 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,5 +1,7 @@
1
1
# collapse 2.1.7
2
2
3
+
* Fixed a bug in `setv()`/`copyv()` where assignments into character (`STRSXP`) or list (`VECSXP`) vectors bypassed R's generational write barrier, writing element pointers directly instead of using `SET_STRING_ELT()`/`SET_VECTOR_ELT()`. This could cause an old-generation target to hold an unrecorded reference to a younger value, which a subsequent young-generation garbage collection could free while still referenced, leading to memory corruption, cryptic `CHAR()`/`SET_STRING_ELT()` errors, or segfaults, most likely under heavy allocation in long-running processes. Thanks @SebKrantz for reporting and diagnosing (#876).
4
+
3
5
* Fixed a bug in `fmatch()` (and thus `%in%`/`%!in%`/`%iin%`/`%!iin%` and joins) where a logical `NA` in `x` could spuriously match a non-`NA` value in `table` (e.g. `2L`) when `table` was not itself logical. Thanks @LJ-Jenkins for reporting (#870).
4
6
5
7
* Fixed a bug in `fslice()` (grouped, `n = 1`, `with.ties = FALSE`) that caused R to crash with a fatal error when a group had only missing values in `order.by`. Thanks @chihyunkim for reporting (#867).
0 commit comments