Codecov Report

Merging #985 (cf45330) into main (38e4b1f) will increase coverage by 1.01%.
The diff coverage is 40.00%.

❗ Current head cf45330 differs from pull request most recent head a94384e. Consider uploading reports for the commit a94384e to get more accurate results

@@            Coverage Diff             @@
##             main     #985      +/-   ##
==========================================
+ Coverage   91.08%   92.10%   +1.01%     
==========================================
  Files          46       46              
  Lines        2691     2646      -45     
==========================================
- Hits         2451     2437      -14     
+ Misses        240      209      -31     

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

Thanks @IndrajeetPatil. Actually we only support the current (4.2), devel and 4 previous versions of R, i.e. 4.1, 4.0, 3.6, 3.5: https://www.tidyverse.org/blog/2019/04/r-version-support/. So I guess instead, the DESCRIPTION should be adapted and if we can simplify parser versions or other things, we should do that I think.

This blog entry mentions that-

our general policy is to support the current version, the devel version, and four previous versions of R.

So I'd assume that the minimum needed R version is now R >= 3.5 across {tidyverse} packages, but that doesn't seem to be the case.

For example:

• For meta-package {tidyverse}, it is R >= 3.3:

https://github.com/tidyverse/tidyverse/blob/3be82832a4982d6d10e25e375cedeb8c3161be35/DESCRIPTION#L17-L18

• For {dplyr}, it is R >= 3.4:

https://github.com/tidyverse/dplyr/blob/73ffdbf29bdb5763b2788c7b41c03a4813c929a8/DESCRIPTION#L30-L31

• For {lubridate}, it is R >= 3.2:

https://github.com/tidyverse/lubridate/blob/0bb49b21c88736240219dc67e7ed0eb3df15d9b1/DESCRIPTION#L32-L34

etc.

But, if you say, you want to support only R >= 3.5, then yes some of the parser-related logic can be simplified.

@lorenzwalthert Let's keep this on hold until R 4.3 release.

Then we can bump the minimum needed R version to 3.6 and get rid of some logic related to parse version 1.

I have created a new PR #986 where I bump the minimum needed R version to 3.5.

Ok, thanks for the analysis. Also, there are some workarounds that can be dropped depending on the R version, mentioned in #581.

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 5035068 is merged into main:

•   :ballot_box_with_check:cache_applying: 38.2ms -> 38.5ms [-2.43%, +3.81%]
•   :ballot_box_with_check:cache_recording: 1.11s -> 1.09s [-3.24%, +0.65%]
•   :ballot_box_with_check:without_cache: 2.76s -> 2.78s [-0.91%, +2.4%]

Further explanation regarding interpretation and methodology can be found in the documentation.

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 33fdf56 is merged into main:

•   :rocket:cache_applying: 28.4ms -> 25.6ms [-12.17%, -7.77%]
•   :ballot_box_with_check:cache_recording: 808ms -> 808ms [-0.49%, +0.46%]
•   :ballot_box_with_check:without_cache: 2.01s -> 2s [-0.64%, +0.02%]

Further explanation regarding interpretation and methodology can be found in the documentation.

The main has deviated far away from this PR, and so I would close this for now.

I will reopen it when R 4.3 release date is announced.