Re: Building of r-cran-prophet failed
On 31 July 2018 at 04:20, Andreas Tille wrote:
| Hi Dirk,
|
| On Mon, Jul 30, 2018 at 06:21:12AM -0500, Dirk Eddelbuettel wrote:
| > | ...
| > | Compiling model (this will take a minute...)
| > | Writing model to: /build/r-cran-prophet-0.3.0.1/debian/r-cran-prophet/usr/lib/R/site-library/prophet/libs
| > | Compiling using binary: /usr/lib/R/bin
| > | Error in rstan::stan_model(stanc_ret = model.stanc, model_name = "prophet_model") :
| > | Boost not found; call install.packages('BH')
| > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
| >
| > Seems BH related.
|
| Obviously. However, I wonder whether the issue is really in prophet or
| rather in rstan. Prophet itself does not have any explicit BH
| dependency but rstan has and the error line seem to indicate that the
| problem is there. I checked rstan test before writing my mail (should
| have probably mentioned this here) which runs without problem.
Well there is a lot more where it could come from:
R> AP <- tools:::CRAN_package_db()
R> tools::package_dependencies(package="prophet", recursive=TRUE, db=AP)
$prophet
[1] "Rcpp" "dplyr" "dygraphs" "extraDistr" "ggplot2" "grid"
[7] "rstan" "scales" "stats" "tidyr" "xts" "assertthat"
[13] "bindrcpp" "glue" "magrittr" "methods" "pkgconfig" "R6"
[19] "rlang" "tibble" "tidyselect" "utils" "BH" "plogr"
[25] "htmlwidgets" "htmltools" "zoo" "digest" "gtable" "lazyeval"
[31] "MASS" "mgcv" "plyr" "reshape2" "viridisLite" "withr"
[37] "StanHeaders" "stats4" "inline" "gridExtra" "RcppEigen" "RColorBrewer"
[43] "dichromat" "munsell" "labeling" "purrr" "stringi" "tools"
[49] "bindr" "grDevices" "graphics" "jsonlite" "yaml" "nlme"
[55] "Matrix" "colorspace" "stringr" "cli" "crayon" "pillar"
[61] "lattice" "fansi" "utf8"
R>
And once you include Suggests and recurse you end up north of 1200 packages:
R> length(tools::package_dependencies(package="prophet", recursive=TRUE, db=AP, \
which=c("Imports", "Depends", "LinkingTo", "Suggests"))[[1]])
[1] 1261
R>
Anyway, you have some time now so I am sure you can dig through this. I was
only tried to help with that suggestion during my morning as your email had
not hint of you being aware of BH.
Dirk
--
http://dirk.eddelbuettel.com | @eddelbuettel | edd@debian.org
Reply to: