Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.
Sign upcreate_codemeta() may lead to 'Error in !auto : invalid argument type' #238
Comments
|
Thanks for the detailed report! Looks like we could probably bootstrap something based on the |
|
Thanks for the bug report! @cboettig I am on maternity leave so can't look into code at the moment. |
|
I proposed a solution to this problem with my PR, this solution is a little bit different from what was done in 'pkgdown', however, it addresses the issue and the function does not break anymore. (Sorry, I realised too late that the new PR stacks with my previous pending PR, I should have better created a new branch from master.) |
* Add tempdir() to avoid to writing guess_other_metadata() into the parent directory. * + swap the default + updating the roxygen docs * + Add myself as a ctb in DESCRIPTION * + mention of changes in NEWS.md + linking this PR * Address #240 and avoid pre-commit hook without user consent * Correct link in NEWS.md #240 * Modify the way the file size is calculated. + Amend guess_fileSize() in guess_other_metadata.R as proposed #239 + Update tests + Drop dependency to 'pkgbuild' + Add NEWS in NEWS.md * Modify the way the file size is calculated. + Amend guess_fileSize() in guess_other_metadata.R as proposed #239 + Update tests + Drop dependency to 'pkgbuild' + Update documentation in 'write_codemeta()' + Upddate corresponding Rd-files + Add NEWS in NEWS.md * Address #238 + Add NEWS * Update tests and add CITATION_ex2 #241
|
closed by #241 |
First thanks for this package, I think it is of great value for the community!
I noticed that
create_codemeta()(path is the package path) stops with throwingThis error is caused by the function
read_citation_with_encoding()(parse_citation.R) which callsutils::readCitationFile(citation_file, meta = meta)and it happens if theCITATIONfile contains the following line:This is nothing special and works on CRAN, since
metacomes with the correct information the moment the package is built on CRAN. I raised a similar issue a while ago in 'pkgdown'(r-lib/pkgdown#493), for which a working solution was found.