-
-
Notifications
You must be signed in to change notification settings - Fork 219
Rcpp 1.0.13 build error with R 4.4.2 (VECTOR_PTR_RO) #1341
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Comments
Hm. Maybe we bit ourselves using 4.4.2 as a cutoff when we meant 4.5.0. |
Yup. I (apparently incorrectly) assumed that R 4.4.2 would've gained those macros. |
FYI, I am seeing this error in a fresh install of R 4.4.2:
|
Yes that is basically the same report as above. If you try the branch @kevinushey just committed via |
I accepted and merged the PR so Updates at r-universe will happen with the hour, that will give you binaries and tarballs too. We will look into CRAN as well. |
I now have rocker/r-base:4.4.2 up and confirmed that current Rcpp at CRAN / github CRAN up to 1.0.13.4 runs into this issue, and 1.0.13.5 passes. Now that I can replicate I can work (tomorrow) on a minimal update 1.0.13-1 that passes. Thanks for everybody's patience while we sort this out. |
@eddelbuettel Thank you! |
- Rcpp 1.0.13 build error with R 4.4.2 (VECTOR_PTR_RO) - RcppCore/Rcpp#1341
I created a new branch release/1.0.13 with two cherry-picked commits: the required 4.4.2 adjustment correcting use of read-only pointers in the C API to R 4.5.0 or later (the issue at heart here) and the switch to Authors@R without which CRAN would not take the package. This content should make it to CRAN as 1.0.13-1. PS It has been shipped, but (as each time) requires a manual approval before even getting to reverse-dependency checks so it may be a few more hours until it is processed. In the meantime, the default branch and its release-candidate version 1.0.13.5 is fine, as are the source and binary packages generated from it at r-universe. PPS We are at the second manual step now as almost invariably given the nearly 3000 reverse dependencies, "something" comes up as a false positive so we are in state 'waiting' and I replied to the email. Normal progress, should hopefully be on CRAN "soon". |
- Rcpp 1.0.13 build error with R 4.4.2 (VECTOR_PTR_RO) - RcppCore/Rcpp#1341
And I just got the 'thanks, on its way to CRAN now' email, this time from Vienna rather than Dortmund. |
Closing this now given the hot-fix is on CRAN as 1.0.13-1. Big thank you to @kyleam for the timely and very focussed bug report, to @RCIIIcm for the confirmation, and of course to @kevinushey for putting PR #1342 up to address it. |
Somehow I get a similar error with |
|
@eddelbuettel Thank you very much! |
…ckage in R version 4.4.2 or higher (RcppCore/Rcpp#1341)
I can successfully build and install Rcpp 1.0.13 with an R built from the current trunk (revision 87283). However, with an R version built from an R-4-4-branch checkout (revision 87276), I get the following error:
9c5f71c (use read-only variants of {STRING/VECTOR}_PTR (#1317), 2024-07-11) added the following conditions:
While it looks like
STRING_PTR_RO
has been around since R 3.5.0,VECTOR_PTR_RO
isn't available untiltrunk@86694
(wch/r-source@85ee9ad34c).VECTOR_PTR_RO
is present in trunk but not the upcoming R 4.4.2 (i.e. the condition above is incorrect).[ +cc @kevinushey ]
The text was updated successfully, but these errors were encountered: