R-xts: update to 0.14.3.
Fixes build with current R.
xts 0.14.3 (2026-09-26)
Fix deprecated special names in structure() calls. (#440)
Fix time-of-day subsetting (e.g. x["T18:00/"]) to create
fractional seconds as decimal fractions instead of raw integers.
The previous behavior could return incorrect results, and error
on R-devel (r90486). (#441)
xts 0.14.2 (2026-02-27)
Removed ATTRIB() and SET_ATTRIB() from C code, since they are
not part of the public R API. (#436)
Fixed crash when endpoints() was called on an xts object with
[10 lines not shown]
R-Rcpp: update to 1.1.2.
Committed during the freeze since it should help with the execinfo
problems on *BSD.
Changes in Rcpp release version 1.1.2 (2026-07-01)
Changes in Rcpp API:
Use of execinfo.h is again conditional to avoid build complexity (Dirk in #1445 addressing #1442)
An internal state component for Datetime is now int (Dirk in #1448 and #1449 fixing #1447)
Three new (in R 4.6.0) attribute accessors are used conditionally (Dirk in #1450 closing #1432)
An UBSAN error in the Sugar-based NA comparison has been corrected (Iñaki in #1453 fixing #1452)
Treatment of Inf outside of integer range in Sugar function has been corrected (Iñaki in #1458 fixing #1455)
[39 lines not shown]
simulavr: fix build with compilers defaulting to c++17
They have std::byte in the namespace (due to 'using namespace std),
which conflicts with the local byte typedef.
Bump PKGREVISION.
R-s2: force C++17
something sets this to build with C++20, but the code is not ready for this
it uses features that were deprecated in C++17 and removed in C++20
Force C++17 to fix the build.
textproc/treemd: update to 0.8.2
(leaf package)
[0.8.2] - 2026-09-13
Fixed
A list nested inside a blockquote or a <details> no longer mispositions what follows it. Its height was measured as one row per item, ignoring both the extra lines a nested item adds to an item's content and the nested blocks the renderer draws underneath it. 1. step holding a fenced block draws four rows and was counted as one, so selecting anything below the list scrolled to the wrong offset, and the error accumulated down the document. A top-level list was already measured correctly by index_elements, which is why this only showed up nested (#88)
Internal
The line counter is now checked against the renderer for every construct. Element line ranges come from the counter and the viewport scrolls to them, so the two disagreeing is what mispositioning is. Two bugs of that shape have shipped, a callout holding a fenced block and now a list holding one, both invisible because nothing compared the numbers. The test measures the same document both ways across twenty constructs and prints any that differ
