[Rcpp-devel] CRAN submission with NOTE about STRING_PTR

[Juan Domingo Esteve]

Dear Patrik and Dirk:

  I have received a few days ago a mail from Prof. Ripley with a similar issue
for my package, scellpam. The error is

checking compiled code ... NOTE
File ‘scellpam/libs/scellpam.so’:
   Found non-API call to R: ‘SET_TYPEOF’

Compiled code should not call non-API entry points in R.

See ‘Writing portable packages’ in the ‘Writing R Extensions’ manual.

which is quite similar to the one received by Patrik Schilter, but in his case with
a different non-API call, 'STRING_PTR'.

I don't know if this may help, but as the answer from Dirk suggests us to make the team
know about this, I write this mail.

  Also, the answer suggests that the issue needs certain time to be fixed.
Might it affect many or all the packages that depends on Rcpp (that, I guess, are
probably a lot of them)?
If this is the case, could we make some kind of collective petition to CRAN
to keep temporarily the packages until this can be fixed? (Because mine, for instance,
according to Prof. Ripley's mail, is scheduled to be retired by next July, 3rd...).

  Thanks a lot for your great effort maintaining Rcpp.

     Juan

El 23/6/24 a las 22:32, Dirk Eddelbuettel escribió:
> 
> Hi Patrik,
> 
> On 23 June 2024 at 21:37, Patrik Schilter wrote:
> | Dear all,
> |
> | First of all: Thank you very much for creating and also maintaining Rcpp so
> | diligently!
> 
> Thanks for the kind words.
> 
> | I am a co-maintainer of the package REndo which depends on Rcpp & RcppEigen.
> | Last week, we submitted an updated versions of our package to CRAN but
> | unfortunately we failed the CRAN incoming checks on
> | `r-devel-linux-x86_64-debian-gcc` and `r-devel-windows-x86_64` with:
> |
> | "Found non-API call to R: ‘STRING_PTR’  Compiled code should not call non-API
> | entry points in R."
> 
> I believe this was a pretty recent addition to the list of symbols flagged.
> And as was mentioned in the github issues you refer to below, all this is
> still very much in flux.
> 
> | The results can be found here:
> | https://urldefense.com/v3/__https://win-builder.r-project.org/incoming_pretest/REndo_2.4.10_20240619_205001__;!!D9dNQwwGXtA!TvAt_hVuu_qlqiOt5YVK6UKTrLg3I16AnoWTMJA4XaJojoTJ7-PrFvFLOUUjM4Nmu6UjeKG-efBk$
> | /Windows/00check.log
> | https://urldefense.com/v3/__https://win-builder.r-project.org/incoming_pretest/REndo_2.4.10_20240619_205001__;!!D9dNQwwGXtA!TvAt_hVuu_qlqiOt5YVK6UKTrLg3I16AnoWTMJA4XaJojoTJ7-PrFvFLOUUjM4Nmu6UjeKG-efBk$
> | /Debian/00check.log
> |
> |
> | From Dirks comment on the Github issue tracker (https://urldefense.com/v3/__https://github.com/RcppCore/__;!!D9dNQwwGXtA!TvAt_hVuu_qlqiOt5YVK6UKTrLg3I16AnoWTMJA4XaJojoTJ7-PrFvFLOUUjM4Nmu6UjeInDQNq2$
> | Rcpp/issues/1311#issuecomment-2185099772), I understand that this issue will
> | not be approached in the upcoming release in July but only in the next release
> | thereafter in January.
> |
> |
> | Two questions:
> |
> | - Does this stem from our usage of `Rcpp::CharacterVector` and could this note
> | be avoided if we did not use `Rcpp::CharacterVector?`
> |
> | - Is my understanding correct, that exiting packages on CRAN which use
> | `Rcpp::CharacterVector` therefore cannot be updated until next year because
> | they will fail the CRAN pre-tests?
> 
> Good questions. But I fear we don't really have answers yet. It took quite a
> bit of work for the just merged PR to improve things relative to use of
> DATAPTR so I fear we will just have to grind it out one issue at time. Any
> and all help is appreciated and welcome so if you learn something in your
> explorations let us know.
> 
> Cheers,  Dirk
> 
> |
> | Thank you very much
> |
> | Patrik
> | _______________________________________________
> | Rcpp-devel mailing list
> | Rcpp-devel at lists.r-forge.r-project.org
> | https://urldefense.com/v3/__https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel__;!!D9dNQwwGXtA!TvAt_hVuu_qlqiOt5YVK6UKTrLg3I16AnoWTMJA4XaJojoTJ7-PrFvFLOUUjM4Nmu6UjeC4ToubL$
> 

-- 
________________________________________________________________
Juan Domingo Esteve
Dept. of Informatics, School of Engineering
University of Valencia
Avda. de la Universidad, s/n.
         46100-Burjasot (Valencia)
            SPAIN

Telephone:      +34-963543572
Fax:            +34-963543550
email:  Juan.Domingo at uv.es
________________________________________________________________