Open richfitz opened 9 years ago
Can duplicate now, working on regression test.
Hmm, this appears to be working for the most part. However, the packaged tar.gz file that R CMD build
creates ends up in a subdirectory, while drat.builder
looks for it at a top-level directory, which causes the drat
step to fail for the package. e.g. see the error resulting from trying to install the ropensci/redland-bindings/R/redland
package here: https://circleci.com/gh/ropensci/drat/97
Weird. Looks like you're handling the subdir logic already here: https://github.com/richfitz/drat.builder/blob/master/R/drat_builder.R#L226-L228 so not sure where this goes wrong. Perhaps it's the use of a sub-subdirectory in the case of redland
, since the other packages with subdirectories seem to be building fine...
I'll try and look at this soon (bother me if I don't). There's a lot of poorly tested book-keeping around the directories, as I was part way through making it possible to dump files elsewhere. But I have little spare time at the moment :(
cool thanks.
No rush; I've added a blacklist to the ropensci drat meanwhile so I can just exclude packages like redland-bindings without having the build break. (Since the packages.txt is generated automatically from the ropkgs API, I need to blacklist rather than whitelist who gets built). Will ping you later.
https://github.com/cboettig/drat/commit/9ee7fedc2009039d602343d08e3e20449d6bc437