i) no := recycling ii) some INTEGER()-etc moved up #3310
Conversation
NEWS.md
Outdated
| @@ -4,6 +4,8 @@ | |||
|
|
|||
| #### NEW FEATURES | |||
|
|
|||
| 1. `:=` no longer recycles too-short RHS vectors to match the LHS length. There was a warning when recycling left a remainder but no warning when the LHS length was an exact multiple of the RHS length (the same behaviour as base R). All recycling is now an error unless `length(RHS)==1` or `length(RHS)==length(LHS)`. Consistent feedback was that recycling is rarely useful and more often a bug. In rare cases where you need to recycle a length>1 vector, use `rep()` explicitly. Single values are still recycled silently as before. | |||
There was a problem hiding this comment.
might be worth doing a revdep run with this commit, i'm curious how many dependencies assume this behavior
There was a problem hiding this comment.
Running. First 80 of 750 complete. Just 1 affected so far: antaresViz
There was a problem hiding this comment.
Complete. 16 affected out of 748. I've updated the top comment.
Codecov Report
@@ Coverage Diff @@
## master #3310 +/- ##
==========================================
+ Coverage 94.8% 94.81% +<.01%
==========================================
Files 65 65
Lines 12094 12104 +10
==========================================
+ Hits 11466 11476 +10
Misses 628 628
Continue to review full report at Codecov.
|
Codecov Report
@@ Coverage Diff @@
## master #3310 +/- ##
=========================================
Coverage ? 94.81%
=========================================
Files ? 65
Lines ? 12104
Branches ? 0
=========================================
Hits ? 11476
Misses ? 628
Partials ? 0
Continue to review full report at Codecov.
|
mattdowle
changed the title
Initially this branch was for #3301. The
memrecycleinternals needed attention for that so I decided to tackle:=recycling at the same time. All the changes to tests look much better without recycling I think.13 CRAN revdeps affected, plus 3 Bioconductor. Hopefully these are mostly unintended bugs that the maintainers will be glad to find.
Output for the 16 out of 748 : fail.log
Tick below means maintainer replied to email saying they agree in principle. After consultation, this PR was merged and #3347 was filed to track PRs we'll submit to those package.
plus 3 out of 152 Bioconductor packages :